//
//  ModelCatalogData.swift
//  Zyquo Agent
//
//  Author: Simon-Pierre Boucher
//  Mail: contact@spboucher.ai
//
//  Built-in model catalog, generated from docs/PROVIDERS.md and the per-provider
//  research files in docs/research/ on 2026-07-30. This file and docs/PROVIDERS.md
//  are a single source of truth and MUST stay in sync: when Phase 7 verification
//  (or any later research pass) changes a model, update both together.
//
//  Scope: chat-completions-capable chat models only. Embeddings, audio/TTS/ASR,
//  realtime, image/video generation, moderation, OCR, robotics, deep-research
//  agents, and Responses-API-only models are excluded by design.
//

import Foundation

enum ModelCatalogData {

    // MARK: - Shared parameter-support presets

    /// OpenAI reasoning models (o-series, gpt-5.x reasoning variants): reject
    /// temperature/top_p/penalties, require max_completion_tokens, accept reasoning_effort.
    private static let openAIReasoning = ParameterSupport(
        temperature: false, topP: false,
        usesMaxCompletionTokens: true, reasoningEffort: true
    )

    // MARK: - OpenAI

    static let openai: [AIModel] = [
        // Flagship GPT-5.6 trio
        AIModel(
            id: "gpt-5.6-sol", provider: .openai, displayName: "GPT-5.6 Sol",
            contextWindow: 1_050_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 30.00),
            parameterSupport: openAIReasoning,
            isRecommended: true
        ),
        AIModel(
            id: "gpt-5.6-terra", provider: .openai, displayName: "GPT-5.6 Terra",
            contextWindow: 1_050_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.50, outputPerMTok: 15.00),
            parameterSupport: openAIReasoning,
            isRecommended: true
        ),
        AIModel(
            id: "gpt-5.6-luna", provider: .openai, displayName: "GPT-5.6 Luna",
            contextWindow: 1_050_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 6.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "chat-latest", provider: .openai, displayName: "ChatGPT Latest",
            contextWindow: 128_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 30.00),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, usesMaxCompletionTokens: true)
        ),
        // Current / recent GPT-5.x
        AIModel(
            id: "gpt-5.5", provider: .openai, displayName: "GPT-5.5",
            contextWindow: 400_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 30.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5.4", provider: .openai, displayName: "GPT-5.4",
            contextWindow: 400_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.50, outputPerMTok: 15.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5.4-mini", provider: .openai, displayName: "GPT-5.4 mini",
            contextWindow: 400_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.75, outputPerMTok: 4.50),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5.4-nano", provider: .openai, displayName: "GPT-5.4 nano",
            contextWindow: 400_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.20, outputPerMTok: 1.25),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5.3-chat-latest", provider: .openai, displayName: "GPT-5.3 Chat Latest",
            contextWindow: 128_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, usesMaxCompletionTokens: true)
        ),
        AIModel(
            id: "gpt-5.2", provider: .openai, displayName: "GPT-5.2",
            contextWindow: 400_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.75, outputPerMTok: 14.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5.2-chat-latest", provider: .openai, displayName: "GPT-5.2 Chat Latest",
            contextWindow: 128_000, maxOutputTokens: 16_000,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.75, outputPerMTok: 14.00),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, usesMaxCompletionTokens: true)
        ),
        AIModel(
            id: "gpt-5.1", provider: .openai, displayName: "GPT-5.1",
            contextWindow: 400_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 10.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5", provider: .openai, displayName: "GPT-5",
            contextWindow: 400_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 10.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5-mini", provider: .openai, displayName: "GPT-5 mini",
            contextWindow: 400_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.25, outputPerMTok: 2.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "gpt-5-nano", provider: .openai, displayName: "GPT-5 nano",
            contextWindow: 400_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.05, outputPerMTok: 0.40),
            parameterSupport: openAIReasoning
        ),
        // o-series reasoning
        AIModel(
            id: "o3", provider: .openai, displayName: "OpenAI o3",
            contextWindow: 200_000, maxOutputTokens: 100_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 8.00),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "o4-mini", provider: .openai, displayName: "OpenAI o4-mini",
            contextWindow: 200_000, maxOutputTokens: 100_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.10, outputPerMTok: 4.40),
            parameterSupport: openAIReasoning
        ),
        AIModel(
            id: "o3-mini", provider: .openai, displayName: "OpenAI o3-mini",
            contextWindow: 200_000, maxOutputTokens: 100_000,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.10, outputPerMTok: 4.40),
            parameterSupport: openAIReasoning,
            isLegacy: true
        ),
        AIModel(
            id: "o1", provider: .openai, displayName: "OpenAI o1",
            contextWindow: 200_000, maxOutputTokens: 100_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 15.00, outputPerMTok: 60.00),
            parameterSupport: openAIReasoning,
            isLegacy: true
        ),
        // Legacy GPT-4.x / 3.5
        AIModel(
            id: "gpt-4.1", provider: .openai, displayName: "GPT-4.1",
            contextWindow: 1_047_576, maxOutputTokens: 32_768,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 8.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-4.1-mini", provider: .openai, displayName: "GPT-4.1 mini",
            contextWindow: 1_047_576, maxOutputTokens: 32_768,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.40, outputPerMTok: 1.60),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-4.1-nano", provider: .openai, displayName: "GPT-4.1 nano",
            contextWindow: 1_047_576, maxOutputTokens: 32_768,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.10, outputPerMTok: 0.40),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-4o", provider: .openai, displayName: "GPT-4o",
            contextWindow: 128_000, maxOutputTokens: 16_384,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.50, outputPerMTok: 10.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-4o-mini", provider: .openai, displayName: "GPT-4o mini",
            contextWindow: 128_000, maxOutputTokens: 16_384,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.15, outputPerMTok: 0.60),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-4-turbo", provider: .openai, displayName: "GPT-4 Turbo",
            contextWindow: 128_000, maxOutputTokens: 4_096,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 10.00, outputPerMTok: 30.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-4", provider: .openai, displayName: "GPT-4",
            contextWindow: 8_192, maxOutputTokens: 8_192,
            capabilities: ModelCapabilities(tools: true),
            pricing: ModelPricing(inputPerMTok: 30.00, outputPerMTok: 60.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "gpt-3.5-turbo", provider: .openai, displayName: "GPT-3.5 Turbo",
            contextWindow: 16_385, maxOutputTokens: 4_096,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.50, outputPerMTok: 1.50),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
    ]

    // MARK: - Anthropic

    static let anthropic: [AIModel] = [
        AIModel(
            id: "claude-opus-5", provider: .anthropic, displayName: "Claude Opus 5",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, thinkingToggle: true),
            isRecommended: true
        ),
        AIModel(
            id: "claude-sonnet-5", provider: .anthropic, displayName: "Claude Sonnet 5",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 3.00, outputPerMTok: 15.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, thinkingToggle: true),
            isRecommended: true
        ),
        AIModel(
            id: "claude-fable-5", provider: .anthropic, displayName: "Claude Fable 5",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 10.00, outputPerMTok: 50.00),
            // Thinking is always on and cannot be disabled — no toggle.
            parameterSupport: ParameterSupport(temperature: false, topP: false, thinkingToggle: false)
        ),
        AIModel(
            id: "claude-opus-4-8", provider: .anthropic, displayName: "Claude Opus 4.8",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, thinkingToggle: true)
        ),
        AIModel(
            id: "claude-opus-4-7", provider: .anthropic, displayName: "Claude Opus 4.7",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, thinkingToggle: true)
        ),
        AIModel(
            id: "claude-opus-4-6", provider: .anthropic, displayName: "Claude Opus 4.6",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport(temperature: true, topP: true, thinkingToggle: true)
        ),
        AIModel(
            id: "claude-sonnet-4-6", provider: .anthropic, displayName: "Claude Sonnet 4.6",
            contextWindow: 1_000_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 3.00, outputPerMTok: 15.00),
            parameterSupport: ParameterSupport(temperature: true, topP: true, thinkingToggle: true)
        ),
        AIModel(
            id: "claude-haiku-4-5-20251001", provider: .anthropic, displayName: "Claude Haiku 4.5",
            contextWindow: 200_000, maxOutputTokens: 64_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 5.00),
            parameterSupport: ParameterSupport(temperature: true, topP: true, thinkingToggle: true)
        ),
        AIModel(
            id: "claude-opus-4-5-20251101", provider: .anthropic, displayName: "Claude Opus 4.5",
            contextWindow: 200_000, maxOutputTokens: 64_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport(temperature: true, topP: true, thinkingToggle: true),
            isLegacy: true
        ),
        AIModel(
            id: "claude-sonnet-4-5-20250929", provider: .anthropic, displayName: "Claude Sonnet 4.5",
            contextWindow: 1_000_000, maxOutputTokens: 64_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 3.00, outputPerMTok: 15.00),
            parameterSupport: ParameterSupport(temperature: true, topP: true, thinkingToggle: true),
            isLegacy: true
        ),
        AIModel(
            id: "claude-opus-4-1-20250805", provider: .anthropic, displayName: "Claude Opus 4.1",
            contextWindow: 200_000, maxOutputTokens: 32_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 15.00, outputPerMTok: 75.00),
            parameterSupport: ParameterSupport(temperature: true, topP: true, thinkingToggle: true),
            isLegacy: true
        ),
    ]

    // MARK: - xAI (Grok)

    static let xai: [AIModel] = [
        AIModel(
            id: "grok-4.5", provider: .xai, displayName: "Grok 4.5",
            contextWindow: 500_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 6.00),
            parameterSupport: ParameterSupport(reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "grok-4.3", provider: .xai, displayName: "Grok 4.3",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 2.50),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "grok-4.20", provider: .xai, displayName: "Grok 4.20 Reasoning",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 2.50),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "grok-4.20-non-reasoning", provider: .xai, displayName: "Grok 4.20 Non-Reasoning",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 2.50),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "grok-code-fast-1", provider: .xai, displayName: "Grok Code Fast 1",
            contextWindow: 256_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 2.00),
            parameterSupport: ParameterSupport(),
            isRecommended: true
        ),
    ]

    // MARK: - Mistral

    static let mistral: [AIModel] = [
        AIModel(
            id: "mistral-medium-latest", provider: .mistral, displayName: "Mistral Medium 3.5",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.50, outputPerMTok: 7.50),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "mistral-large-latest", provider: .mistral, displayName: "Mistral Large 3",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.50, outputPerMTok: 1.50),
            parameterSupport: .openAIDefault,
            isRecommended: true
        ),
        AIModel(
            id: "mistral-small-latest", provider: .mistral, displayName: "Mistral Small 4",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.15, outputPerMTok: 0.60),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "codestral-latest", provider: .mistral, displayName: "Codestral",
            contextWindow: 256_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 0.90),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "ministral-14b-latest", provider: .mistral, displayName: "Ministral 3 14B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.20, outputPerMTok: 0.20),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "ministral-8b-latest", provider: .mistral, displayName: "Ministral 3 8B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.15, outputPerMTok: 0.15),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "ministral-3b-latest", provider: .mistral, displayName: "Ministral 3 3B",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.10, outputPerMTok: 0.10),
            parameterSupport: .openAIDefault
        ),
        // Legacy / deprecated (still served; hidden by default)
        AIModel(
            id: "magistral-medium-latest", provider: .mistral, displayName: "Magistral Medium",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 5.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "devstral-latest", provider: .mistral, displayName: "Devstral 2",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.40, outputPerMTok: 2.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "open-mistral-nemo", provider: .mistral, displayName: "Mistral Nemo",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.15, outputPerMTok: 0.15),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
    ]

    // MARK: - Google Gemini

    static let gemini: [AIModel] = [
        AIModel(
            id: "gemini-3.6-flash", provider: .gemini, displayName: "Gemini 3.6 Flash",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.50, outputPerMTok: 7.50),
            parameterSupport: ParameterSupport(reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "gemini-3.5-flash", provider: .gemini, displayName: "Gemini 3.5 Flash",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.50, outputPerMTok: 9.00),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemini-3.5-flash-lite", provider: .gemini, displayName: "Gemini 3.5 Flash-Lite",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 2.50),
            parameterSupport: ParameterSupport(reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "gemini-3.1-pro-preview", provider: .gemini, displayName: "Gemini 3.1 Pro (Preview)",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 12.00),
            parameterSupport: ParameterSupport(reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "gemini-3.1-flash-lite", provider: .gemini, displayName: "Gemini 3.1 Flash-Lite",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.25, outputPerMTok: 1.50),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemini-2.5-pro", provider: .gemini, displayName: "Gemini 2.5 Pro",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 10.00),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemini-2.5-flash", provider: .gemini, displayName: "Gemini 2.5 Flash",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 2.50),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemini-2.5-flash-lite", provider: .gemini, displayName: "Gemini 2.5 Flash-Lite",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.10, outputPerMTok: 0.40),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        // Rolling aliases (auto-track the latest release; pricing varies with target)
        AIModel(
            id: "gemini-pro-latest", provider: .gemini, displayName: "Gemini Pro (Latest)",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemini-flash-latest", provider: .gemini, displayName: "Gemini Flash (Latest)",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemini-flash-lite-latest", provider: .gemini, displayName: "Gemini Flash-Lite (Latest)",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        // Preview / secondary
        AIModel(
            id: "gemini-3-flash-preview", provider: .gemini, displayName: "Gemini 3 Flash (Preview)",
            contextWindow: 1_048_576, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.50, outputPerMTok: 3.00),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "gemma-4-26b-a4b-it", provider: .gemini, displayName: "Gemma 4 26B",
            contextWindow: 262_144, maxOutputTokens: 32_768,
            capabilities: ModelCapabilities(jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "gemma-4-31b-it", provider: .gemini, displayName: "Gemma 4 31B",
            contextWindow: 262_144, maxOutputTokens: 32_768,
            capabilities: ModelCapabilities(jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
    ]

    // MARK: - Alibaba Qwen (DashScope)

    static let qwen: [AIModel] = [
        // Flagship commercial
        AIModel(
            id: "qwen3.7-max", provider: .qwen, displayName: "Qwen3.7 Max",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.50, outputPerMTok: 7.50),
            parameterSupport: ParameterSupport(thinkingToggle: true),
            isRecommended: true
        ),
        AIModel(
            id: "qwen3.7-plus", provider: .qwen, displayName: "Qwen3.7 Plus",
            contextWindow: 1_000_000, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.32, outputPerMTok: 1.28),
            parameterSupport: ParameterSupport(thinkingToggle: true),
            isRecommended: true
        ),
        AIModel(
            id: "qwen3.7-flash", provider: .qwen, displayName: "Qwen3.7 Flash",
            contextWindow: 1_000_000, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.03, outputPerMTok: 0.13),
            parameterSupport: ParameterSupport(thinkingToggle: true),
            isRecommended: true
        ),
        AIModel(
            id: "qwen3.6-plus", provider: .qwen, displayName: "Qwen3.6 Plus",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3.6-flash", provider: .qwen, displayName: "Qwen3.6 Flash",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3.5-plus", provider: .qwen, displayName: "Qwen3.5 Plus",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3.5-flash", provider: .qwen, displayName: "Qwen3.5 Flash",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        // Stable aliases (previous-gen commercial)
        AIModel(
            id: "qwen-max", provider: .qwen, displayName: "Qwen Max",
            contextWindow: 128_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen-plus", provider: .qwen, displayName: "Qwen Plus",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen-turbo", provider: .qwen, displayName: "Qwen Turbo",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true),
            isLegacy: true
        ),
        AIModel(
            id: "qwen-flash", provider: .qwen, displayName: "Qwen Flash",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        // Coder family
        AIModel(
            id: "qwen3-coder-plus", provider: .qwen, displayName: "Qwen3 Coder Plus",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-coder-flash", provider: .qwen, displayName: "Qwen3 Coder Flash",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-coder-next", provider: .qwen, displayName: "Qwen3 Coder Next",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-coder-480b-a35b-instruct", provider: .qwen, displayName: "Qwen3 Coder 480B A35B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        // Vision-language
        AIModel(
            id: "qwen3-vl-plus", provider: .qwen, displayName: "Qwen3 VL Plus",
            contextWindow: 1_000_000, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3-vl-flash", provider: .qwen, displayName: "Qwen3 VL Flash",
            contextWindow: 1_000_000, maxOutputTokens: 65_536,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3-vl-235b-a22b-instruct", provider: .qwen, displayName: "Qwen3 VL 235B Instruct",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-vl-235b-a22b-thinking", provider: .qwen, displayName: "Qwen3 VL 235B Thinking",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qvq-max", provider: .qwen, displayName: "QVQ Max",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(requiresStreaming: true)
        ),
        // Reasoning-only
        AIModel(
            id: "qwq-plus", provider: .qwen, displayName: "QwQ Plus",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(requiresStreaming: true)
        ),
        // Open-weights Qwen hosted on DashScope
        AIModel(
            id: "qwen3.5-397b-a17b", provider: .qwen, displayName: "Qwen3.5 397B A17B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3.5-122b-a10b", provider: .qwen, displayName: "Qwen3.5 122B A10B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3.5-35b-a3b", provider: .qwen, displayName: "Qwen3.5 35B A3B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "qwen3-235b-a22b-instruct-2507", provider: .qwen, displayName: "Qwen3 235B Instruct 2507",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-235b-a22b-thinking-2507", provider: .qwen, displayName: "Qwen3 235B Thinking 2507",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-next-80b-a3b-instruct", provider: .qwen, displayName: "Qwen3 Next 80B Instruct",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "qwen3-next-80b-a3b-thinking", provider: .qwen, displayName: "Qwen3 Next 80B Thinking",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport()
        ),
        // Third-party models hosted on DashScope
        AIModel(
            id: "deepseek-v4-pro", provider: .qwen, displayName: "DeepSeek V4 Pro (DashScope)",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "deepseek-v4-flash", provider: .qwen, displayName: "DeepSeek V4 Flash (DashScope)",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "glm-5.2", provider: .qwen, displayName: "GLM 5.2 (DashScope)",
            contextWindow: 198_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
        AIModel(
            id: "kimi-k2.7-code", provider: .qwen, displayName: "Kimi K2.7 Code (DashScope)",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: nil,
            parameterSupport: ParameterSupport(thinkingToggle: true)
        ),
    ]

    // MARK: - DeepSeek

    static let deepseek: [AIModel] = [
        AIModel(
            id: "deepseek-v4-flash", provider: .deepseek, displayName: "DeepSeek V4 Flash",
            contextWindow: 1_000_000, maxOutputTokens: 384_000,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.14, outputPerMTok: 0.28),
            parameterSupport: ParameterSupport(reasoningEffort: true, thinkingToggle: true),
            isRecommended: true
        ),
        AIModel(
            id: "deepseek-v4-pro", provider: .deepseek, displayName: "DeepSeek V4 Pro",
            contextWindow: 1_000_000, maxOutputTokens: 384_000,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.435, outputPerMTok: 0.87),
            parameterSupport: ParameterSupport(reasoningEffort: true, thinkingToggle: true),
            isRecommended: true
        ),
    ]

    // MARK: - Kimi (Moonshot AI)

    static let kimi: [AIModel] = [
        AIModel(
            id: "kimi-k3", provider: .kimi, displayName: "Kimi K3",
            contextWindow: 1_048_576, maxOutputTokens: 131_072,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 3.00, outputPerMTok: 15.00),
            // Thinking always on with preserved thinking; depth via reasoning_effort.
            parameterSupport: ParameterSupport(temperature: false, topP: false, usesMaxCompletionTokens: true, reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "kimi-k2.7-code", provider: .kimi, displayName: "Kimi K2.7 Code",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.95, outputPerMTok: 4.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, usesMaxCompletionTokens: true),
            isRecommended: true
        ),
        AIModel(
            id: "kimi-k2.7-code-highspeed", provider: .kimi, displayName: "Kimi K2.7 Code Highspeed",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.90, outputPerMTok: 8.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, usesMaxCompletionTokens: true)
        ),
        AIModel(
            id: "kimi-k2.6", provider: .kimi, displayName: "Kimi K2.6",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.95, outputPerMTok: 4.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, usesMaxCompletionTokens: true, thinkingToggle: true)
        ),
        AIModel(
            id: "kimi-k2.5", provider: .kimi, displayName: "Kimi K2.5",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.60, outputPerMTok: 3.00),
            parameterSupport: ParameterSupport(temperature: false, topP: false, usesMaxCompletionTokens: true, thinkingToggle: true)
        ),
        // Legacy moonshot-v1 "classic" series (temperature capped at 1.0)
        AIModel(
            id: "moonshot-v1-8k", provider: .kimi, displayName: "Moonshot v1 8K",
            contextWindow: 8_192, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.20, outputPerMTok: 2.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "moonshot-v1-32k", provider: .kimi, displayName: "Moonshot v1 32K",
            contextWindow: 32_768, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 3.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "moonshot-v1-128k", provider: .kimi, displayName: "Moonshot v1 128K",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 5.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "moonshot-v1-auto", provider: .kimi, displayName: "Moonshot v1 Auto",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: nil,
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "moonshot-v1-8k-vision-preview", provider: .kimi, displayName: "Moonshot v1 8K Vision",
            contextWindow: 8_192, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.20, outputPerMTok: 2.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "moonshot-v1-32k-vision-preview", provider: .kimi, displayName: "Moonshot v1 32K Vision",
            contextWindow: 32_768, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 3.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
        AIModel(
            id: "moonshot-v1-128k-vision-preview", provider: .kimi, displayName: "Moonshot v1 128K Vision",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 5.00),
            parameterSupport: .openAIDefault,
            isLegacy: true
        ),
    ]

    // MARK: - Perplexity

    static let perplexity: [AIModel] = [
        AIModel(
            id: "sonar", provider: .perplexity, displayName: "Sonar",
            contextWindow: 128_000, maxOutputTokens: 128_000,
            capabilities: ModelCapabilities(jsonMode: true, citations: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 1.00),
            parameterSupport: ParameterSupport(),
            isRecommended: true
        ),
        AIModel(
            id: "sonar-pro", provider: .perplexity, displayName: "Sonar Pro",
            contextWindow: 200_000, maxOutputTokens: 8_000,
            capabilities: ModelCapabilities(jsonMode: true, citations: true),
            pricing: ModelPricing(inputPerMTok: 3.00, outputPerMTok: 15.00),
            parameterSupport: ParameterSupport(),
            isRecommended: true
        ),
        AIModel(
            id: "sonar-reasoning-pro", provider: .perplexity, displayName: "Sonar Reasoning Pro",
            contextWindow: 128_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(reasoning: true, jsonMode: true, citations: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 8.00),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
        AIModel(
            id: "sonar-deep-research", provider: .perplexity, displayName: "Sonar Deep Research",
            contextWindow: 128_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(reasoning: true, jsonMode: true, citations: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 8.00),
            parameterSupport: ParameterSupport(reasoningEffort: true)
        ),
    ]

    // MARK: - Together AI

    static let together: [AIModel] = [
        AIModel(
            id: "moonshotai/Kimi-K3", provider: .together, displayName: "Kimi K3",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 3.00, outputPerMTok: 15.00),
            parameterSupport: .openAIDefault,
            isRecommended: true
        ),
        AIModel(
            id: "moonshotai/Kimi-K2.7-Code", provider: .together, displayName: "Kimi K2.7 Code",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.95, outputPerMTok: 4.00),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "moonshotai/Kimi-K2.6", provider: .together, displayName: "Kimi K2.6",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.20, outputPerMTok: 4.50),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "deepseek-ai/DeepSeek-V4-Pro", provider: .together, displayName: "DeepSeek V4 Pro",
            contextWindow: 512_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.74, outputPerMTok: 3.48),
            parameterSupport: .openAIDefault,
            isRecommended: true
        ),
        AIModel(
            id: "zai-org/GLM-5.2", provider: .together, displayName: "GLM 5.2",
            contextWindow: 512_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.40, outputPerMTok: 4.40),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3.7-Max", provider: .together, displayName: "Qwen3.7 Max",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 3.75),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, requiresStreaming: true)
        ),
        AIModel(
            id: "Qwen/Qwen3.7-Plus", provider: .together, displayName: "Qwen3.7 Plus",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.32, outputPerMTok: 1.28),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, requiresStreaming: true)
        ),
        AIModel(
            id: "Qwen/Qwen3.6-Plus", provider: .together, displayName: "Qwen3.6 Plus",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.50, outputPerMTok: 3.00),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, requiresStreaming: true)
        ),
        AIModel(
            id: "Qwen/Qwen3.5-9B", provider: .together, displayName: "Qwen3.5 9B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.17, outputPerMTok: 0.25),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, requiresStreaming: true)
        ),
        AIModel(
            id: "meta-llama/Llama-3.3-70B-Instruct-Turbo", provider: .together, displayName: "Llama 3.3 70B Turbo",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.04, outputPerMTok: 1.04),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "openai/gpt-oss-120b", provider: .together, displayName: "GPT-OSS 120B",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.15, outputPerMTok: 0.60),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "openai/gpt-oss-20b", provider: .together, displayName: "GPT-OSS 20B",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.05, outputPerMTok: 0.20),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, reasoningEffort: true)
        ),
        AIModel(
            id: "nvidia/nemotron-3-ultra-550b-a55b", provider: .together, displayName: "Nemotron 3 Ultra 550B",
            contextWindow: 512_288, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.60, outputPerMTok: 3.60),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "MiniMaxAI/MiniMax-M3", provider: .together, displayName: "MiniMax M3",
            contextWindow: 524_288, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 1.20),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            // vision disabled 2026-07-30: Together's endpoint accepts image
            // input but streams an empty answer (verified live) — text only.
            id: "google/gemma-4-31B-it", provider: .together, displayName: "Gemma 4 31B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: false, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.39, outputPerMTok: 0.97),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, requiresStreaming: true)
        ),
        AIModel(
            id: "thinkingmachines/Inkling", provider: .together, displayName: "Inkling",
            contextWindow: 524_288, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 4.05),
            parameterSupport: .openAIDefault
        ),
    ]

    // MARK: - DeepInfra

    static let deepinfra: [AIModel] = [
        // Proxied frontier models (Claude / Gemini under DeepInfra billing)
        AIModel(
            id: "anthropic/claude-fable-5", provider: .deepinfra, displayName: "Claude Fable 5",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 10.00, outputPerMTok: 50.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "anthropic/claude-opus-5", provider: .deepinfra, displayName: "Claude Opus 5",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "anthropic/claude-sonnet-5", provider: .deepinfra, displayName: "Claude Sonnet 5",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 10.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "anthropic/claude-opus-4-8", provider: .deepinfra, displayName: "Claude Opus 4.8",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 5.00, outputPerMTok: 25.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "anthropic/claude-haiku-4-5", provider: .deepinfra, displayName: "Claude Haiku 4.5",
            contextWindow: 200_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.00, outputPerMTok: 5.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "google/gemini-3.1-pro", provider: .deepinfra, displayName: "Gemini 3.1 Pro",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.00, outputPerMTok: 12.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "google/gemini-3.5-flash", provider: .deepinfra, displayName: "Gemini 3.5 Flash",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.50, outputPerMTok: 9.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "google/gemini-3.1-flash-lite", provider: .deepinfra, displayName: "Gemini 3.1 Flash-Lite",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.25, outputPerMTok: 1.50),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "google/gemini-2.5-pro", provider: .deepinfra, displayName: "Gemini 2.5 Pro",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.25, outputPerMTok: 10.00),
            parameterSupport: ParameterSupport()
        ),
        AIModel(
            id: "google/gemini-2.5-flash", provider: .deepinfra, displayName: "Gemini 2.5 Flash",
            contextWindow: 1_000_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 2.50),
            parameterSupport: ParameterSupport()
        ),
        // Open-weight chat models
        AIModel(
            id: "deepseek-ai/DeepSeek-V4-Pro", provider: .deepinfra, displayName: "DeepSeek V4 Pro",
            contextWindow: 1_048_576, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 1.30, outputPerMTok: 2.60),
            parameterSupport: .openAIDefault,
            isRecommended: true
        ),
        AIModel(
            id: "deepseek-ai/DeepSeek-V4-Flash", provider: .deepinfra, displayName: "DeepSeek V4 Flash",
            contextWindow: 1_048_576, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.09, outputPerMTok: 0.18),
            parameterSupport: .openAIDefault,
            isRecommended: true
        ),
        AIModel(
            id: "deepseek-ai/DeepSeek-V3.1", provider: .deepinfra, displayName: "DeepSeek V3.1",
            contextWindow: 163_840, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.25, outputPerMTok: 0.95),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "deepseek-ai/DeepSeek-R1-0528", provider: .deepinfra, displayName: "DeepSeek R1 0528",
            contextWindow: 163_840, maxOutputTokens: nil,
            capabilities: ModelCapabilities(reasoning: true),
            pricing: ModelPricing(inputPerMTok: 0.50, outputPerMTok: 2.15),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "moonshotai/Kimi-K2.7-Code", provider: .deepinfra, displayName: "Kimi K2.7 Code",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.74, outputPerMTok: 3.50),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "moonshotai/Kimi-K2.6", provider: .deepinfra, displayName: "Kimi K2.6",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.75, outputPerMTok: 3.50),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "moonshotai/Kimi-K2.5", provider: .deepinfra, displayName: "Kimi K2.5",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.45, outputPerMTok: 2.25),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, requiresStreaming: true)
        ),
        AIModel(
            id: "zai-org/GLM-5.2", provider: .deepinfra, displayName: "GLM 5.2",
            contextWindow: 1_048_576, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.75, outputPerMTok: 2.40),
            parameterSupport: .openAIDefault,
            isRecommended: true
        ),
        AIModel(
            id: "zai-org/GLM-4.7", provider: .deepinfra, displayName: "GLM 4.7",
            contextWindow: 202_752, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.40, outputPerMTok: 1.75),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3.7-Max", provider: .deepinfra, displayName: "Qwen3.7 Max",
            contextWindow: 256_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.50, outputPerMTok: 7.50),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3.5-397B-A17B", provider: .deepinfra, displayName: "Qwen3.5 397B A17B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.45, outputPerMTok: 3.00),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3-235B-A22B-Instruct-2507", provider: .deepinfra, displayName: "Qwen3 235B Instruct 2507",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.09, outputPerMTok: 0.55),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3-235B-A22B-Thinking-2507", provider: .deepinfra, displayName: "Qwen3 235B Thinking 2507",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.23, outputPerMTok: 2.30),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3-Coder-480B-A35B-Instruct-Turbo", provider: .deepinfra, displayName: "Qwen3 Coder 480B Turbo",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 1.00),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "Qwen/Qwen3-VL-235B-A22B-Instruct", provider: .deepinfra, displayName: "Qwen3 VL 235B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.20, outputPerMTok: 0.88),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8", provider: .deepinfra, displayName: "Llama 4 Maverick",
            contextWindow: 1_048_576, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.20, outputPerMTok: 0.80),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "meta-llama/Llama-4-Scout-17B-16E-Instruct", provider: .deepinfra, displayName: "Llama 4 Scout",
            contextWindow: 327_680, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.10, outputPerMTok: 0.30),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "meta-llama/Llama-3.3-70B-Instruct-Turbo", provider: .deepinfra, displayName: "Llama 3.3 70B Turbo",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.10, outputPerMTok: 0.32),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo", provider: .deepinfra, displayName: "Llama 3.1 8B Turbo",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.02, outputPerMTok: 0.04),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "openai/gpt-oss-120b", provider: .deepinfra, displayName: "GPT-OSS 120B",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.037, outputPerMTok: 0.17),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "openai/gpt-oss-20b", provider: .deepinfra, displayName: "GPT-OSS 20B",
            contextWindow: 131_072, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.03, outputPerMTok: 0.14),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, reasoningEffort: true)
        ),
        AIModel(
            id: "MiniMaxAI/MiniMax-M3", provider: .deepinfra, displayName: "MiniMax M3",
            contextWindow: 524_288, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.30, outputPerMTok: 1.20),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B", provider: .deepinfra, displayName: "Nemotron 3 Ultra 550B",
            contextWindow: 262_144, maxOutputTokens: nil,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.50, outputPerMTok: 2.20),
            parameterSupport: .openAIDefault
        ),
        AIModel(
            id: "mistralai/Mistral-Small-3.2-24B-Instruct-2506", provider: .deepinfra, displayName: "Mistral Small 3.2 24B",
            contextWindow: 128_000, maxOutputTokens: nil,
            capabilities: ModelCapabilities(vision: true, tools: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.075, outputPerMTok: 0.20),
            parameterSupport: .openAIDefault
        ),
        // google/gemma-4-31B-it removed 2026-07-30: endpoint hangs (60s+, zero
        // bytes) on chat completions — see docs/PROVIDERS.md Phase 7 amendments.
    ]

    // MARK: - Cerebras

    static let cerebras: [AIModel] = [
        AIModel(
            id: "gpt-oss-120b", provider: .cerebras, displayName: "GPT-OSS 120B",
            contextWindow: 131_072, maxOutputTokens: 40_000,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.35, outputPerMTok: 0.75),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, usesMaxCompletionTokens: true, reasoningEffort: true),
            isRecommended: true
        ),
        AIModel(
            id: "gemma-4-31b", provider: .cerebras, displayName: "Gemma 4 31B",
            contextWindow: 131_072, maxOutputTokens: 40_000,
            capabilities: ModelCapabilities(vision: true, tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 0.99, outputPerMTok: 1.49),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, usesMaxCompletionTokens: true, reasoningEffort: true)
        ),
        AIModel(
            id: "zai-glm-4.7", provider: .cerebras, displayName: "GLM 4.7",
            contextWindow: 131_072, maxOutputTokens: 40_000,
            capabilities: ModelCapabilities(tools: true, reasoning: true, jsonMode: true),
            pricing: ModelPricing(inputPerMTok: 2.25, outputPerMTok: 2.75),
            parameterSupport: ParameterSupport(frequencyPenalty: true, presencePenalty: true, usesMaxCompletionTokens: true, reasoningEffort: true),
            // Scheduled for discontinuation on 2026-08-17.
            isLegacy: true
        ),
    ]

    // MARK: - All providers

    static let all: [AIModel] =
        openai + anthropic + xai + mistral + gemini + qwen +
        deepseek + kimi + perplexity + together + deepinfra + cerebras
}