GPQA Diamond (OpenRouter)
?what it ishighest score on this metric 96.5%
90º / 135 · OpenRouter Benchmarks, latest version we checked
Qwen3 235B A22B Thinking 2507
Benchmarks, price, context and versions
Also searched as
Qwen3-235B-A22B-Thinking-2507
Direct answer: Qwen3 235B A22B Thinking 2507 is a model from Alibaba. 79.1% on GPQA Diamond (OpenRouter), per OpenRouter Benchmarks. On OpenRouter, US$ 0.45 per 1 million input tokens and US$ 3.50 per 1 million output tokens.
✓ Good for
- running on your own infrastructure (open weights)
✕ Not the best for
- technical reasoning as the model appears through the API (#90 of 135 on GPQA Diamond (OpenRouter))
- agents that follow rules and use tools (#98 of 131 on τ-bench Verified (airline))
This block follows a rule of this section, using source data: if the model is in the top 25% of the metric, the item goes under "Good for"; if it is in the bottom half, it goes under "Not the best for". The middle band appears in neither. Price and context use OpenRouter; scores use Epoch AI or OpenRouter Benchmarks, in the latest version we checked.
Benchmarks
τ-bench Verified (airline)
?highest score on this metric 82%
98º / 131 · OpenRouter Benchmarks, latest version we checked
Key specs
- Lab
- Alibaba
- Origin
- China
- Family
- Qwen Thinking provisional data
- Weights?
- open
- License
- apache-2.0
- Input US$/1M?
- US$ 0.45
- Output US$/1M
- US$ 3.50
- Context?
- 128,000 tokens
- Parameters
- 235,093,634,560
- Input
- text
- OpenRouter usage?
- no data
- Downloads (30 days)?
- 8,177
- Dates in sources
- 2025, listed on OpenRouter
- 2025, Hugging Face repo created
Version history
- Qwen3-VL-2B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-2B-Thinking Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-32B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-32B-Thinking Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-8B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-4B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-8B-Thinking Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-4B-Thinking Hugging Face Hub, Hugging Face repo created
- Qwen: Qwen3 VL 30B A3B Thinking OpenRouter, listed on OpenRouter
- Qwen3-VL-30B-A3B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-235B-A22B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-30B-A3B-Thinking Hugging Face Hub, Hugging Face repo created
- Qwen3-Next-80B-A3B-Thinking-FP8 Hugging Face Hub, Hugging Face repo created
- Qwen3-VL-235B-A22B-Thinking Hugging Face Hub, Hugging Face repo created
Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch.
FAQ
What is Qwen3 235B A22B Thinking 2507?
Qwen3 235B A22B Thinking 2507 is a model from Alibaba, with open weights, used for text, agents.
How much does Qwen3 235B A22B Thinking 2507 cost?
On OpenRouter, US$ 0.45 per 1 million input tokens and US$ 3.50 per 1 million output tokens.
What is the context window of Qwen3 235B A22B Thinking 2507?
128,000 tokens, per OpenRouter.
Is Qwen3 235B A22B Thinking 2507 open source?
Yes, it has open weights. License: apache-2.0.
What is Qwen3 235B A22B Thinking 2507's GPQA Diamond (OpenRouter) score?
79.1% on GPQA Diamond (OpenRouter), per OpenRouter Benchmarks.
Is Qwen3 235B A22B Thinking 2507 good for coding?
No data: there is no SWE-bench Verified measurement for it in open sources.
Who made Qwen3 235B A22B Thinking 2507?
It was made by Alibaba, a lab from China.
provisional data Questions under review: Semrush had no API balance during collection. Answers use source data only.
Models from Alibaba
Sources
- Hugging Face Hub, metadados públicos do Hub; licença de cada modelo no repositório, latest version we checked.
- OpenRouter, dados públicos da API; exports de benchmarks e rankings em CC BY 4.0, latest version we checked.
- OpenRouter Benchmarks, CC BY 4.0, latest version we checked.
Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.