Qwen3-235B-A22B-Thinking (Jul 2025)
ECI#85
144Qwen3-235B-A22B-Thinking (Jul 2025) is an open-weights model from Alibaba, released July 25, 2025. It has an ECI score of 144 and ranks 85th of 249 tracked models. At the time of its release, it ranked 7th of 151 tracked models. At $0.23/Mtok for input tokens and $2.30/Mtok for output tokens, it is more expensive per token than similarly-capable models.
Input price$0.23 / 1M tokens
Output price$2.30 / 1M tokens
Parameters2.4e11
Training compute4.8e24 FLOP
Qwen3-235B-A22B-Thinking (Jul 2025) vs select alternatives
Reasoning
| Domain | Benchmark | Qwen3-235B-A22B-Thinking (Jul 2025) | |||
|---|---|---|---|---|---|
ECI scoreBest score 166 | 144 | 139 | 144 | 145 | |
| Ranking | #85 | #115 | #83 | #76 | |
| Agent | Vending-Bench 2Best score 11181.9 | -11.3 | — | — | 573.6 |
| Games | Chess PuzzlesBest score 72.0% | ||||
Mystery Game PuzzlesBest score 84.0% | |||||
| Long context | Fiction.liveBenchBest score 97.2% | ||||
| Mathematics | OTIS Mock AIME 2024-2025Best score 100.0% | ||||
| Science | CritPtBest score 32.3% | ||||
SciCodeBest score 62.0% | |||||
GPQA DiamondBest score 95.8% | |||||
| Software engineering | WeirdML (v2)Best score 92.9% |
