Qwen3 235B A22B Thinking 2507
ModelOpen sourceActiveby Alibaba (Qwen) · family “qwen3b-a22b-thinking”
Qwen3-235B-A22B-Thinking-2507 is a high-performance, open-weight Mixture-of-Experts (MoE) language model optimized for complex reasoning tasks. It activates 22B of its 235B parameters per forward pass and natively supports up to 262,144... (description from the OpenRouter listing)
in: textout: textReasoningTool useStructured output
Epoch Capabilities Index
Not scored by Epoch AI
Released
25 Jul 2025
Context window
131K tokens
Max output
118K tokens
Input $ / 1M tokens
$0.23
Output $ / 1M tokens
$2.30
Cached input $ / 1M
Unknown
Each benchmark shown separately with its own source
Benchmark results (3)
| Benchmark | Domain | Score | vs best recorded | Setting | Run | Source |
|---|---|---|---|---|---|---|
| FrontierMath-Tier-4-2025-07-01-Privatesuperseded | math | 0.0% | 0% | — | 11 Dec 2025 | Epoch ↗ |
| FrontierMath-2025-02-28-Privatesuperseded | math | 8.5% ±1.7 | 16% | — | 11 Dec 2025 | Epoch ↗ |
| WeirdML | coding | 38.7% ±0.0 | 41% | — | — | External ↗ |
Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.
API list price over time
Price history
$ per 1M tokens
13 recorded prices on 1 Oct 2025 (OpenRouter listing + Internet Archive snapshots); re-read on every data refresh, most recently 8h ago. Steps show when the price changed.
Events
No events recorded yet.