Updated daily at 05:00 UTC
Vision-language
o4-mini
- Released
- 16 Apr 2025
- Status
- Deprecated
- Retired
- 23 Oct 2026
- Context
- 200K
- Max output
- 100K
- Input per 1M tokens
- $1.10
- Output per 1M tokens
- $4.40
- Open weights
- No
- License
- proprietary
- Parameters
- —
- Knowledge cutoff
- Jun 2024
- Reasoning
- Yes
- Modalities
- text, image → text
Category scores
Out of 100, from independent and third-party results only, by method category-2026-09-29.1.
| Category | Score | Rank | Coverage |
|---|---|---|---|
| Knowledge & reasoning | 26.6 | 64 | 3 of 4 |
| Coding & software engineering | 37.8 | Not ranked | 2 of 5 |
| Vision & multimodal understanding | 50.1 | 6 | 2 of 3 |
| Factuality & hallucination | 19.6 | 59 | 1 of 1 |
| Human preference | 36.4 | Not ranked | 2 of 5 |
Benchmark results
Every result we hold for this model, with who ran it and where it was published.
Knowledge & reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Chess Puzzles | Puzzles solved | 26.0% | Independent | high | 8 Dec 2025 | Epoch AI |
| Chess Puzzles | Puzzles solved | 20.0% | Independent | medium | 7 Aug 2026 | Epoch AI |
| Chess Puzzles | Puzzles solved | 14.0% | Independent | low | 11 Jul 2026 | Epoch AI |
| DTBench (Decision Theory Benchmark) | Accuracy (capability questions) | 77.6% | Third party | high | — | https://conceptualreasoning.ai/dtbench via Epoch AI |
| LMCA (Language Model Conceptual Argumentation) | Correlation with expert ratings (Pearson r) | 0.265 | Third party | high | — | https://conceptualreasoning.ai/lmca via Epoch AI |
| Mystery Game Puzzles | Puzzles solved | 5.0% | Independent | high | 27 Aug 2026 | Epoch AI |
| SimpleBench | Score (AVG@5) | 38.7% | Third party | high | — | SimpleBench Leaderboard via Epoch AI |
Scientific reasoning
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GPQA Diamond | Accuracy | 79.6% | Independent | high | 16 Apr 2025 | Epoch AI |
| GPQA Diamond | Accuracy | 77.8% | Independent | medium | 7 Aug 2026 | Epoch AI |
| GPQA Diamond | Accuracy | 75.3% | Independent | low | 11 Jul 2026 | Epoch AI |
Mathematics
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| FrontierMath Tier 4 | Accuracy | 4.9% | Independent | high | 11 Jun 2026 | Epoch AI |
| FrontierMath Tier 4 (v1) | Accuracy | 6.3% | Independent | high | 1 Jul 2025 | Epoch AI |
| FrontierMath Tier 4 (v1) | Accuracy | 2.1% | Independent | medium | 7 Aug 2025 | Epoch AI |
| FrontierMath Tiers 1-3 | Accuracy | 36.1% | Independent | high | 11 Jun 2026 | Epoch AI |
| FrontierMath Tiers 1-3 | Accuracy | 28.8% | Independent | medium | 27 Aug 2026 | Epoch AI |
| FrontierMath Tiers 1-3 | Accuracy | 16.1% | Independent | low | 27 Aug 2026 | Epoch AI |
| FrontierMath Tiers 1-3 (v1) | Accuracy | 24.8% | Independent | high | 13 Nov 2025 | Epoch AI |
| FrontierMath Tiers 1-3 (v1) | Accuracy | 19.0% | Independent | medium | 13 Nov 2025 | Epoch AI |
| FrontierMath Tiers 1-3 (v1) | Accuracy | 10.7% | Independent | low | 16 Nov 2025 | Epoch AI |
| MATH Level 5 | Accuracy | 97.8% | Independent | high | 16 Apr 2025 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 81.7% | Independent | high | 16 Apr 2025 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 73.3% | Independent | medium | 7 Aug 2026 | Epoch AI |
| OTIS Mock AIME 2024-2025 | Accuracy | 57.8% | Independent | low | 13 Jul 2026 | Epoch AI |
Coding & software engineering
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Aider Polyglot | Percent correct | 72.0% | Third party | high | — | https://aider.chat/docs/leaderboards/ via Epoch AI |
| CadEval | Overall pass rate | 62.0% | Third party | medium | — | CadEval Dashboard via Epoch AI |
| GSO | OPT@1 | 3.6% | Third party | high | — | GSO Leaderboard via Epoch AI |
| METR Time Horizons | Average task success (Epoch's column) | 63.9% | Third party | medium | — | METR - Measuring AI Ability to Complete Long Tasks via Epoch AI |
| WeirdML (v2) | Average accuracy | 52.6% | Third party | high | — | https://htihle.github.io/weirdml.html via Epoch AI |
Long context
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Fiction.LiveBench | % correct at 16K tokens (Epoch's column) | 77.8% | Third party | medium | — | Fiction.live leaderboard via Epoch AI |
Vision & multimodal understanding
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| GeoBench | Country accuracy | 64.0% | Third party | high | — | GeoBench leaderboard via Epoch AI |
| GeoBench | Country accuracy | 64.0% | Third party | medium | — | GeoBench leaderboard via Epoch AI |
| Visual Physics Comprehension Test (VPCT) | Correct | 57.5% | Third party | medium | — | VPCT leaderboard via Epoch AI |
Factuality & hallucination
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| SimpleQA Verified | Accuracy (% correct) | 19.6% | Independent | low | 27 Aug 2026 | Epoch AI |
| SimpleQA Verified | Accuracy (% correct) | 18.8% | Independent | high | 27 Aug 2026 | Epoch AI |
Writing
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| LLM Creative Story-Writing Benchmark V1 (Lech Mazur) | Mean rubric score (percent of the 10-point scale) | 75.0% | Third party | medium | — | lechmazur/writing Github repository via Epoch AI |
Human preference
| Benchmark | Metric | Value | Provenance | Setting | Evaluated | Source |
|---|---|---|---|---|---|---|
| Text Arena (arena.ai) | Arena score (Elo scale) | 1391 | Independent | — | 25 Sep 2026 | Arena |
| Text Arena (arena.ai) | Votes | 44637 | Independent | — | 25 Sep 2026 | Arena |
| Vision Arena (arena.ai) | Arena score (Elo scale) | 1201 | Independent | — | 28 Sep 2026 | Arena |
| Vision Arena (arena.ai) | Votes | 41354 | Independent | — | 28 Sep 2026 | Arena |
Other o4 versions
- o4-mini-deep-research24 Jun 2025