Qwen3-235B-A22B-Thinking (Jul 2025)
Qwen3-235B-A22B-Thinking (Jul 2025) (Alibaba) was released on Jul 25, 2025. In the Epoch AI capability ranking (ECI), it ranks #102 out of 267 models, with 143.81. On Arena, which measures human preference in blind comparisons, it has 1,384 ELO points (#188). Against models of the same generation, its best result is on AIME 2024 and 2025 (OTIS): 86.7%, #3 of 53, tied with 1 others.
Where Qwen3-235B-A22B-Thinking (Jul 2025) stands out and where it falls behind, by the numbers
Qwen3-235B-A22B-Thinking (Jul 2025) has published results on 16 tests compiled by Epoch AI. The yardstick here is its generation: its position among the models released between Oct 28, 2024 and Oct 23, 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 16 tests, one by one
| Test | Score | In its generation | Overall | Test record |
|---|---|---|---|---|
| GPQA Diamond | 80.1% | #6 / 53 | #83 / 173 | 96.4% Gemini 4 Argon |
| AIME 2024 and 2025 (OTIS) | 86.7% | #3 / 53 (tied with 1) | #66 / 168 | 100.0% GPT-5.5 |
| Dtbench | 80.3% | #12 / 46 | #70 / 148 | 98.9% Claude Opus 5.5 |
| Lmca | 29.31 | #10 / 38 | #75 / 121 | 68.23 Claude Opus 5.5 |
| WeirdML | 41.0% | #21 / 38 | #70 / 112 | 93.6% GPT-6 Astra |
| SimpleQA Verified | 40.4% | #5 / 14 | #47 / 81 | 76.0% Gemini 4 Argon |
| Game puzzles | 9.0% | #6 / 12 (tied with 1) | #64 / 77 | 84.0% GPT-6 Astra |
| SimpleBench | 31.0% | #18 / 27 | #56 / 74 | 81.9% Claude Fable 5 |
| MATH level 5 | 68.9% | #29 / 37 | #31 / 73 | 98.1% GPT-5 |
| FrontierMath | 8.5% | #10 / 30 | #38 / 65 | 51.7% GPT-5.5 |
| ForecastBench | 59.70 | #17 / 24 | #32 / 62 | 62.50 o3 |
| Vending Bench 2 | -11.34 | #6 / 8 | #54 / 57 | 15,515 GPT-6 Astra |
| FrontierMath tier 4 | 0.0% | #15 / 18 (tied with 3) | #40 / 48 | 39.6% GPT-5.5 Pro |
| Aider Polyglot | 59.60 | #14 / 31 (tied with 1) | #15 / 38 | 88.00 GPT-5 |
| Creative writing (Lech Mazur) | 8.30 | #8 / 32 (tied with 1) | #9 / 38 | 8.60 GPT-5 |
| Fiction.liveBench | 68.8% | #7 / 30 (tied with 1) | #8 / 31 | 100.0% o3 |
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 Qwen3-235B-A22B-Thinking (Jul 2025) 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 |
|---|---|---|---|---|---|
| Gemini 3.1 Flash-Lite | 144.47 | 1,460 | n/a | Mar 3, 2026 | |
| Gemini 2.5 Pro Exp (Mar 2025) | Google DeepMind | 144.20 | n/a | n/a | Mar 25, 2025 |
| Grok 4 Fast | xAI | 144.19 | 1,437 | n/a | Sep 19, 2025 |
| Claude Opus 4.1 | Anthropic | 144.09 | 1,479 | n/a | Aug 5, 2025 |
| Qwen3-235B-A22B-Thinking (Jul 2025) | Alibaba | 143.81 | 1,384 | n/a | Jul 25, 2025 |
| GLM-4.7 | Z.ai (Zhipu AI) | 143.79 | 1,495 | n/a | Dec 22, 2025 |
| Gemini 2.5 Flash (Sep 2025) | Google DeepMind | 143.00 | 1,450 | n/a | Sep 25, 2025 |
Qwen3-235B-A22B-Thinking (Jul 2025) or GLM-4.7?
GLM-4.7 is the model from another lab closest to Qwen3-235B-A22B-Thinking (Jul 2025) in capability. Across the 7 tests both took, this is the score:
| Test | Qwen3-235B-A22B-Thinking (Jul 2025) | GLM-4.7 |
|---|---|---|
| Vending Bench 2 | -11.34 | 2,377 |
| FrontierMath | 8.5% | 2.4% |
| SimpleBench | 31.0% | 47.7% |
| SimpleQA Verified | 40.4% | 32.2% |
| GPQA Diamond | 80.1% | 83.3% |
| AIME 2024 and 2025 (OTIS) | 86.7% | 83.3% |
| FrontierMath tier 4 | 0.0% | 0.0% |
Unweighted score: each test counts one point, whether math or coding. The dot marks who did better on each row.
Other models from Alibaba
The models from the same lab closest in time, newest first.
| Model | Release | ECI | ELO | LiveBench | Access |
|---|---|---|---|---|---|
| Qwen3-Max-Instruct | Sep 24, 2025 | 142.46 | n/a | n/a | API access |
| Qwen3-Max | Sep 24, 2025 | 142.34 | 1,478 | n/a | API access |
| Qwen3-30B-A3B-Thinking (Jul 2025) | Jul 30, 2025 | 139.61 | 1,347 | n/a | Open weights (unrestricted) |
| Qwen3-30B-A3B-Instruct (Jul 2025) | Jul 29, 2025 | 137.42 | n/a | n/a | Open weights (unrestricted) |
| Qwen3-235B-A22B-Thinking-2507 | Jul 25, 2025 | 143.88 | 1,483 | n/a | Open weights (unrestricted) |
| Qwen3-235B-A22B-Thinking (Jul 2025) | Jul 25, 2025 | 143.81 | 1,384 | n/a | Open weights (unrestricted) |
| Qwen3-235B-A22B-Instruct (Jul 2025) | Jul 25, 2025 | 138.92 | n/a | n/a | Open weights (unrestricted) |
| Qwen3-32B | Apr 29, 2025 | 138.49 | 1,362 | n/a | Open weights (unrestricted) |
| Qwen3-14B | Apr 29, 2025 | 138.22 | n/a | n/a | Open weights (unrestricted) |
Have you used Qwen3-235B-A22B-Thinking (Jul 2025)? 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 Qwen3-235B-A22B-Thinking (Jul 2025)
Where does Qwen3-235B-A22B-Thinking (Jul 2025) rank among AI models?
In the ROO3 AI Benchmark, ordered by the Epoch AI ECI, Qwen3-235B-A22B-Thinking (Jul 2025) ranks #102 out of 267 models, with an ECI of 143.81. 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 Qwen3-235B-A22B-Thinking (Jul 2025) better than GLM-4.7?
Across the 7 tests both models took and Epoch AI compiled, Qwen3-235B-A22B-Thinking (Jul 2025) comes out ahead on 3, behind on 3 and tied on 1. 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 Qwen3-235B-A22B-Thinking (Jul 2025) 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 Qwen3-235B-A22B-Thinking (Jul 2025) are: AIME 2024 and 2025 (OTIS) (#3 of 53); GPQA Diamond (#6 of 53); Fiction.liveBench (#7 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