spb/ai-atlas

1# GPT-4.123> For the complete documentation index, see [llms.txt](/llms.txt). Markdown versions of documentation pages are available by appending `.md` to the page URL.45> Smartest non-reasoning model67Model ID: `gpt-4.1`89GPT-4.1 excels at instruction following and tool calling, with broad 10knowledge across domains. It features a 1M token context window, and11low latency without a reasoning step.1213Note that we recommend starting with [GPT-5](/api/docs/models/gpt-5) for complex tasks.1415## Model details1617- Default snapshot: `gpt-4.1-2025-04-14`18- Input modalities: text, image19- Output modalities: text20- 1,047,576 context window21- 32,768 max output tokens22- Jun 01, 2024 knowledge cutoff2324## Pricing2526Pricing is based on the number of tokens used, or other metrics based on the model type. For tool-specific models, like search and computer use, there’s a fee per tool call. See details in the [pricing page](/api/docs/pricing).2728### Text tokens2930| Metric | Price | Unit |31| --- | ---: | --- |32| Input | $2 | 1M tokens |33| Cached input | $0.5 | 1M tokens |34| Output | $8 | 1M tokens |3536## Endpoints3738| Endpoint | Route | Support |39| --- | --- | --- |40| Live | `v1/live/sessions` | Not supported |41| Chat Completions | `v1/chat/completions` | Supported |42| Responses | `v1/responses` | Supported |43| Realtime | `v1/realtime` | Not supported |44| Realtime translation | `v1/realtime/translations` | Not supported |45| Realtime transcription | `v1/realtime/transcription_sessions` | Not supported |46| Assistants | `v1/assistants` | Supported |47| Batch | `v1/batch` | Supported |48| Fine-tuning | `v1/fine-tuning` | Supported |49| Embeddings | `v1/embeddings` | Not supported |50| Image generation | `v1/images/generations` | Not supported |51| Videos | `v1/videos` | Not supported |52| Image edit | `v1/images/edits` | Not supported |53| Speech generation | `v1/audio/speech` | Not supported |54| Transcription | `v1/audio/transcriptions` | Not supported |55| Translation | `v1/audio/translations` | Not supported |56| Moderation | `v1/moderations` | Not supported |57| Completions (legacy) | `v1/completions` | Not supported |5859## Supported features6061- streaming62- structured_outputs63- predicted_outputs64- function_calling65- file_search66- file_uploads67- image_input68- web_search69- fine_tuning70- prompt_caching7172## Supported tools7374Tools supported by this model when using the Responses API.7576- function_calling77- web_search78- file_search79- image_generation80- code_interpreter81- mcp8283## Quick comparison8485| Model | Input | Cached input | Output |86| --- | ---: | ---: | ---: |87| GPT-4.1 | $2 | $0.5 | $8 |88| GPT-4o | $2.5 | $1.25 | $10 |89| o3-mini | $1.1 | $0.55 | $4.4 |9091## Snapshots9293Snapshots let you lock in a specific version of the model so that performance and behavior remain consistent. Below is a list of all available snapshots and aliases for GPT-4.1.9495- `gpt-4.1-2025-04-14`9697## Rate limits9899Rate limits ensure fair and reliable access to the API by placing specific caps on requests, tokens, audio duration, or other usage within a given time period. Your usage tier determines how high these limits are set and automatically increases as you send more requests and spend more on the API.100101### default102103| Tier | RPM | TPM | Batch queue limit |104| --- | ---: | ---: | ---: |105| Tier 1 | 500 | 30,000 | 90,000 |106| Tier 2 | 5,000 | 450,000 | 1,350,000 |107| Tier 3 | 5,000 | 800,000 | 50,000,000 |108| Tier 4 | 10,000 | 2,000,000 | 200,000,000 |109| Tier 5 | 10,000 | 30,000,000 | 5,000,000,000 |110111### Long Context112113> 128k input tokens114115| Tier | RPM | TPM | Batch queue limit |116| --- | ---: | ---: | ---: |117| Tier 1 | 100 | 200,000 | 2,000,000 |118| Tier 2 | 250 | 500,000 | 20,000,000 |119| Tier 3 | 500 | 1,000,000 | 40,000,000 |120| Tier 4 | 1,000 | 5,000,000 | 100,000,000 |121| Tier 5 | 4,000 | 10,000,000 | 1,000,000,000 |122