o3 / o4-mini
OpenAIOpenAI's o3 and o4-mini are reasoning models that spend an extended chain of thought before answering: o3 is the full-scale model and o4-mini the cost-efficient one. Both were offered in ChatGPT and through OpenAI's API, without downloadable weights, and both have since been retired from ChatGPT. OpenAI lists GPT-5 as the successor to o3.
Scored as the o3 and o4-mini pair; the related o3-mini is not scored separately.
Openness
1 high confidence- weights
- closed
- data
- closed
- code
- closed
- license
- Proprietary(API+ChatGPT only)
Proprietary OpenAI reasoning models; no weights, data, or code released.
- https://en.wikipedia.org/wiki/OpenAI_o3 recorded 2026-09-29
Article on the o3 family, describing o3-mini as a "specialized alternative" to o1 for "technical domains requiring precision and speed" and the models as offered through ChatGPT and the API. Searched the body for open-source, open weights, weights, license and code release: none.
- https://platform.openai.com/docs/models/o3 recorded 2026-09-29
o3 API model page: "o3 is a well-rounded and powerful model across domains". Searched the body for weights, open-weight, open source and license: none; the model is offered only as a hosted API endpoint.
Adoption
5 medium confidenceo3 and o4-mini were offered across ChatGPT tiers as well as the API, and OpenAI publishes no per-model usage count, so the reading was taken from the ChatGPT surface rather than a standalone figure. Both have since been retired from ChatGPT; o3 remains available in the API.
- https://en.wikipedia.org/wiki/OpenAI_o3 recorded 2026-08-13
"OpenAI released o3-mini to all ChatGPT users (including free-tier) and some API users"; o3 and o3-pro on the paid tiers; no standalone per-model user count published
Capability
4 high confidenceOpenAI reported 87.7% on GPQA Diamond and 71.7% on SWE-bench Verified for o3 at its April 2025 launch, results that placed it at the frontier for reasoning models at the time. OpenAI now lists GPT-5 as its successor, so the family sits a step behind the current frontier.
- https://cdn.openai.com/pdf/2221c875-02dc-4789-800b-e7758f3722c1/o3-and-o4-mini-system-card.pdf recorded 2026-08-13
primary system card, still resolving; models described as combining "state-of-the-art reasoning with full tool capabilities"
- https://en.wikipedia.org/wiki/OpenAI_o3 recorded 2026-08-13
"o3 achieved a score of 87.7% on the GPQA Diamond benchmark"; "On SWE-bench Verified ... o3 scored 71.7%, compared to 48.9% for o1. On Codeforces, o3 reached an Elo score of 2727, whereas o1 scored 1891". No o4-mini AIME figure on the page.
Verified 2026-09-29, except some scores, still awaiting re-confirmation