xAI (Grok) API — atlas navigation
Status: provider added 2026-09-18/19; catalogue, pricing, rate limits, deprecations, management/account endpoints, auth/headers/errors, SDKs and gRPC surface LIVE_VERIFIED/DOCUMENTED by the models agent; other areas are written by their own agents (links below are stable even if a page is not present yet).
Sources: https://docs.x.ai/developers/** (182 pages, sources/xai/pages/**, retrieved 2026-09-19), OpenAPI sources/xai/openapi/openapi.json (38 paths), xai-sdk 1.19.0, live probes logged in reports/live-requests.jsonl (provider: xai).
Last verified: 2026-09-18.
Essentials
| Item | Value |
|---|---|
| Base URL | https://api.x.ai/v1 (US-only: https://us.api.x.ai/v1, grok-4.6, ×1.1; undocumented eu-west-1.api.x.ai serves grok-4.3; mTLS mtls.api.x.ai; gRPC api.x.ai:443) |
| Auth | Authorization: Bearer xai-… — no version/beta headers |
| Management API | https://management-api.x.ai with a separate Management key (ACCOUNT_RESTRICTED for this atlas) |
| Cheapest text model | grok-4.3 ($1.25 / $2.50 per 1M; reasoning_effort: none → no reasoning tokens); recommended model grok-4.6 ($2 / $6) |
| Live catalogue | 12 ids (GET /v1/models, 2026-09-19): 7 language, 3 image, 2 video |
| Exact cost | usage.cost_in_usd_ticks (1 USD = 10^10 ticks) on every inference response |
Pages
Models, pricing, limits, lifecycle (this agent)
- Models catalogue + capability / endpoint / tool matrices —
generated/fragments/models/xai-models.json - Pricing —
generated/fragments/pricing/xai-pricing.json - Rate limits —
generated/fragments/rate-limits/xai-rate-limits.json - Deprecations, retirements and release notes —
generated/fragments/deprecations/xai-deprecations.json - Management API + account/catalogue endpoints —
generated/fragments/endpoints/xai-models-management.json - Authentication, base URLs, headers, errors —
generated/fragments/headers/xai-headers.json,generated/fragments/errors/xai-errors.json - SDKs and client compatibility —
generated/fragments/sdks/xai-sdks.json - gRPC API surface
Text inference (other agents)
- Chat Completions
- Responses API
- Anthropic-compatible Messages (
/v1/messages,/v1/complete) - Streaming (SSE + WebSocket Responses mode)
- Structured outputs
- Reasoning (
reasoning_effort, encrypted reasoning, multi-agent) - Deferred completions
- Batches
Tools
Files, collections, media, voice
- Files API
- Collections
- Images (Imagine image generation / editing)
- Videos (Imagine video generation / editing / extension)
- Voice (speech-to-speech realtime, STT, TTS)
Cross-provider
- Comparisons · Errors · Endpoints · Architecture · Security
Scripts
scripts/discover_xai_models.py— re-run catalogue discovery (/v1/models+ typed lists) with diff.scripts/gen_xai_fragments.py— rebuild models / pricing / rate-limit fragments from the raws + documented facts.source scripts/lib.sh; xai GET /v1/models·python3 scripts/live.py xai GET /v1/me.