As of March 11, 2025, we’ve released the building blocks of our new Agents platform. For details, see our API docs for our Responses API, Tools including Web Search, File Search, and Computer Use, and our Agents SDK with Tracing.
⚠️ This article is about the state of access to our reasoning models in the API and does not cover access to these models in ChatGPT.
Model Access by Usage Tier
o1, o1-preview, o1-mini, o1-pro
Access: All API users on tiers 1 through 5.
Endpoints: Chat Completions, Assistants API, Batch API.
o1-prois more compute-intensive and provides stronger performance. Pricing details are available in the model reference.
o3-mini
Access: All API users on tiers 1 through 5.
Capabilities: Function calling, Structured Outputs, developer messages, reasoning effort, streaming.
Endpoints: Chat Completions, Assistants API, Batch API.
o3
Access: Tiers 4 and 5, with some exceptions.
Note: Users on Tiers 1–3 can gain access by completing organization verification.
Endpoint: Responses API.
o4-mini
Access: All API users on tiers 1 through 5, subject to organization verification.
Endpoints: Responses API.
Special Capability: This is the only model that supports Reinforcement Fine-Tuning (RFT), which allows you to optimize model behavior using a custom reward signal.
How do I find my usage tier?
Navigate to the Limits menu in the API Platform dashboard to see your usage tier and what models you have access to.