Back to Gap Map Model components / Fine-tuned / chat models

o3 / o4-mini

closed / Overall score: n/a

OpenAI's o3 and o4-mini are reasoning models that work through an extended chain of thought before answering: o3 is the full-scale model and o4-mini the cost-efficient one. Both are still served through OpenAI's API, without downloadable weights, but are deprecated there, with GPT-5 and GPT-5 Mini named as their successors; o4-mini is scheduled for removal on October 23, 2026 and o3 on December 11, 2026. Both have left ChatGPT.

Scored as the o3 and o4-mini pair; the related o3-mini is not scored separately.

Openness

1 high confidence

1.0

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

not assessed

o3 and o4-mini are still served through OpenAI's API, where both are marked deprecated, but OpenAI publishes no usage count for either, and both have left ChatGPT, so ChatGPT's user figures no longer describe them. No level is assigned until a count of their own use appears.

https://developers.openai.com/api/docs/deprecations recorded 2026-10-09
OpenAI's deprecations page schedules "October 23, 2026 o4-mini-2025-04-16 | o4-mini gpt-5.6-terra" and "Dec 11, 2026 o3-2025-04-16 gpt-5.6-sol". The page carries no usage figure.
https://developers.openai.com/api/docs/models/o3 recorded 2026-10-09
o3 is marked "Deprecated" and described as "Reasoning model for complex tasks, succeeded by GPT-5"; the page still offers the o3 alias and the o3-2025-04-16 snapshot. Searched the body for users, requests and any usage count: none.
https://developers.openai.com/api/docs/models/o4-mini recorded 2026-10-09
o4-mini is marked "Deprecated" and described as "Fast, cost-efficient reasoning model, succeeded by GPT-5 Mini"; the page still offers the o4-mini alias and snapshot. Searched the body for users, requests and any usage count: none.

Capability

3 high confidence

3.0

o3 and o4-mini led OpenAI's reasoning line in 2025 and are still served through the API, but newer models have overtaken them. On the Artificial Analysis index o3 now scores below the median for comparable models and about ten points under Gemini 3.1 Pro, a step below the group of models just behind the frontier leaders.

https://artificialanalysis.ai/leaderboards/models recorded 2026-10-09
The leaderboard's embedded model records, ranked by intelligenceIndex, put Claude Opus 5.5 (Max) first at 57.62, GPT-6 Astra (Max) at 52.67, Gemini 3.1 Pro Preview at 29.72, o3 at 20.20, o4-mini (High) at 16.65, Llama 4 Maverick at 9.99 and GPT-4o (Nov '24) at 8.44. The digest drifts between fetches because the page carries live pricing and throughput figures.
https://artificialanalysis.ai/models/o3 recorded 2026-10-09
"o3 scores 20 on the Artificial Analysis Intelligence Index, placing it below average among comparable models (median: 26)"; Intelligence rank "# 157 / 226".
https://artificialanalysis.ai/models/o4-mini recorded 2026-10-09
"o4-mini (High) scores 17 on the Artificial Analysis Intelligence Index, placing it at the lower end among comparable models (median: 26)"; Intelligence rank "# 170 / 226"; the page marks the model deprecated in favor of GPT-5 mini.

Verified 2026-09-29