Qwen3-235B-A22B-Thinking-2507
Qwen3-235B-A22B-Thinking-2507 is an AI model published by 阿里巴巴, released on 2025-07-25, for 推理大模型, with 2350.0B parameters, and 256K tokens context length, requiring about 470.77 GB storage, under the Apache 2.0 license.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Qwen3-235B-A22B-Thinking-2507 currently shows benchmark results led by MMLU Pro (26 / 116, score 84.40), Creative Writing (5 / 22, score 86.10), LiveCodeBench (32 / 109, score 74.10). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.
| Variant name | Version type | Quantization | Model size | HuggingFace link |
|---|---|---|---|---|
| Qwen3-235B-A22B-Thinking-2507-FP8ℹ️ | Instruct | FP8 | 236.45 GB | Download link |
阿里巴巴开源的Qwen3-235B-A22B模型的升级版本，最早的Qwen3-235B-A22B模型是在2025年4月28日随着Qwen3系列一起发布，当时是推理和非推理模式混合的架构模型，后来阿里发现这个模式不好，因此在2025年7月份发布了更新版的模型，即不支持推理模式的Qwen3-235B-A22B-2507和支持推理模式的Qwen3-235B-A22B-Thinking-2507。
Qwen3-235B-A22B-Thinking-2507最多可以支持80K的推理过程长度，最高支持32K的答案输出，是当前推理过程最长的模型之一！
欢迎关注 DataLearner 官方微信，获得最新 AI 技术推送
| Modality | Input | Output |
|---|---|---|
| Text | $0.7 | $8.4 |
