gpt-oss-120b
gpt-oss-120b (OpenAI) was released on Aug 5, 2025. In the Epoch AI capability ranking (ECI), it ranks #124 out of 267 models, with 140.00. On Arena, which measures human preference in blind comparisons, it has 1,385 ELO points (#191). Against models of the same generation, its best result is on AIME 2024 and 2025 (OTIS): 88.9%, #2 of 53.
Where gpt-oss-120b stands out and where it falls behind, by the numbers
gpt-oss-120b has published results on 18 tests compiled by Epoch AI. The yardstick here is its generation: its position among the models released between Nov 8, 2024 and Nov 3, 2025 that took the same test. Against the whole archive, which includes 2023 models, a recent release lands near the top of almost everything, and that says nothing.
Where it stands out
Tests where it sits in the top third of its own generation.
Where it falls behind
Tests where it sits in the bottom half of its own generation.
See all 18 tests, one by one
| Test | Score | In its generation | Overall | Test record |
|---|---|---|---|---|
| GPQA Diamond | 75.8% | #16 / 53 | #95 / 173 | 96.4% Gemini 4 Argon |
| AIME 2024 and 2025 (OTIS) | 88.9% | #2 / 53 | #59 / 168 | 100.0% GPT-5.5 |
| Dtbench | 76.3% | #18 / 46 | #83 / 148 | 98.9% Claude Opus 5.5 |
| Lmca | 22.14 | #21 / 38 | #90 / 121 | 68.23 Claude Opus 5.5 |
| CritPt | 1.1% | #6 / 30 (tied with 1) | #74 / 112 | 32.3% GPT-5.6 Sol |
| WeirdML | 48.2% | #8 / 38 | #50 / 112 | 93.6% GPT-6 Astra |
| SciCode | 34.0% | #15 / 26 | #89 / 104 | 66.9% Claude Opus 5.5 |
| ALE-Bench | 576 | #13 / 18 | #73 / 82 | 2,951 GPT-6 Astra |
| Game puzzles | 0.0% | #12 / 12 | #77 / 77 | 84.0% GPT-6 Astra |
| SimpleBench | 22.1% | #26 / 27 | #69 / 74 | 81.9% Claude Fable 5 |
| Vending Bench 2 | -21.53 | #7 / 8 | #55 / 57 | 15,515 GPT-6 Astra |
| Apex Agents | 4.4% | #1 / 1 | #50 / 50 | 82.4% Gemini 4 Argon |
| Aider Polyglot | 41.80 | #22 / 31 | #25 / 38 | 88.00 GPT-5 |
| Creative writing (Lech Mazur) | 7.71 | #15 / 32 | #18 / 38 | 8.60 GPT-5 |
| Terminal-Bench | 18.7% | #9 / 11 | #30 / 33 | 84.7% GPT-5.5 |
| METR: time horizon | 41.98 | #10 / 14 | #18 / 29 | 719 Claude Opus 4.6 |
| Surface Evolver Bench | 25.0% | #1 / 1 | #24 / 24 | 95.0% Claude Fable 5 |
| Algotune | 1.41 | #6 / 9 | #12 / 15 | 2.05 GPT-5.2 |
For each test, the model's best effort level counts, as in the ranking. Source of the scores: Epoch AI, under CC BY 4.0.
What to compare gpt-oss-120b with
The models from other labs closest in capability and released in the same period, at most two per lab. Click any of them to see its page.
| Model | Lab | ECI | ELO | LiveBench | Release |
|---|---|---|---|---|---|
| GLM-4.6 | Z.ai (Zhipu AI) | 140.78 | 1,500 | n/a | Sep 30, 2025 |
| Grok-3 mini | xAI | 140.33 | 1,394 | n/a | Jun 24, 2025 |
| Kimi K2 Instruct | Moonshot | 140.13 | n/a | n/a | Jul 12, 2025 |
| gpt-oss-120b | OpenAI | 140.00 | 1,385 | n/a | Aug 5, 2025 |
| DeepSeek-V3.1 | DeepSeek | 139.91 | 1,469 | n/a | Aug 21, 2025 |
| Qwen3-30B-A3B-Thinking (Jul 2025) | Alibaba | 139.61 | 1,347 | n/a | Jul 30, 2025 |
| Qwen3.5-9B | Alibaba | 139.43 | n/a | n/a | Feb 24, 2026 |
gpt-oss-120b or Grok-3 mini?
Grok-3 mini is the model from another lab closest to gpt-oss-120b in capability. Across the 5 tests both took, this is the score:
| Test | gpt-oss-120b | Grok-3 mini |
|---|---|---|
| Aider Polyglot | 41.80 | 49.30 |
| AIME 2024 and 2025 (OTIS) | 88.9% | 77.8% |
| WeirdML | 48.2% | 42.6% |
| Creative writing (Lech Mazur) | 7.71 | 7.35 |
| GPQA Diamond | 75.8% | 76.3% |
Unweighted score: each test counts one point, whether math or coding. The dot marks who did better on each row.
Other models from OpenAI
The models from the same lab closest in time, newest first.
| Model | Release | ECI | ELO | LiveBench | Access |
|---|---|---|---|---|---|
| GPT-5.1 | Nov 13, 2025 | 149.76 | 1,495 | n/a | API access |
| GPT-5 Pro | Oct 7, 2025 | 150.74 | n/a | n/a | API access |
| GPT-5 | Aug 7, 2025 | 150.00 | 1,436 | n/a | API access |
| GPT-5 mini | Aug 7, 2025 | 145.43 | 1,406 | n/a | API access |
| GPT-5 nano | Aug 7, 2025 | 139.33 | 1,356 | n/a | API access |
| gpt-oss-120b | Aug 5, 2025 | 140.00 | 1,385 | n/a | Open weights (unrestricted) |
| gpt-oss-20b | Aug 5, 2025 | 137.81 | 1,315 | n/a | Open weights (unrestricted) |
| o3-pro | Jun 10, 2025 | 147.37 | n/a | n/a | API access |
| o3 | Apr 16, 2025 | 146.87 | n/a | n/a | API access |
Have you used gpt-oss-120b? Rate it
Users' opinion, not measurement: the rating is kept apart from the ranking and does not move anyone's position. One vote per browser, no sign-up, and you can change your rating later.
What people ask about gpt-oss-120b
Where does gpt-oss-120b rank among AI models?
In the ROO3 AI Benchmark, ordered by the Epoch AI ECI, gpt-oss-120b ranks #124 out of 267 models, with an ECI of 140.00. The ECI is a composite index that combines several hard benchmarks into a single scale. The position changes when a new model comes out, and the page is refreshed six times a day.
Is gpt-oss-120b better than Grok-3 mini?
Across the 5 tests both models took and Epoch AI compiled, gpt-oss-120b comes out ahead on 3, behind on 2 and tied on 0. It is a test-by-test count with no weighting: a win in math counts the same as a win in coding. To choose between the two, look at the tests that resemble your own use.
What is gpt-oss-120b best at?
Against models of the same generation (released up to 9 months before and up to 3 months after it), the best relative results of gpt-oss-120b are: AIME 2024 and 2025 (OTIS) (#2 of 53); WeirdML (#8 of 38); CritPt (#6 of 30). The yardstick is its position among the models of that period that took the same test.
Where do the numbers on this page come from?
From the same three sources as the ranking, all openly licensed: Epoch AI (ECI and individual tests, CC BY 4.0), Arena (human-preference ELO, CC BY 4.0) and LiveBench (overall average, CC BY-SA 4.0). ROO3 does not measure any model. When the page shows a figure published by the maker itself, it appears in the written analysis, labeled as the maker's, and never enters the ranking.
Where each number comes from
Epoch AI, "AI Benchmarking Hub". Published at epoch.ai. Data under the CC BY 4.0 licence, reordered by roo3.co.
The Arena (LMArena) leaderboard, from the public lmarena-ai/leaderboard-dataset on Hugging Face, under the CC BY 4.0 licence. We are not affiliated with the Arena.
LiveBench, by the LiveBench team, published at livebench.ai under the CC BY-SA 4.0 license. The overall average is recomputed by roo3.co with the same formula as the website (average of the categories), and those numbers stay under CC BY-SA 4.0. We are not affiliated with LiveBench.
← See the full ranking