Qwen3 235B A22B Instruct 2507
Qwen3-235B-A22B-Instruct-2507 is a multilingual, instruction-tuned mixture-of-experts language model based on the Qwen3-235B architecture, with 22B active parameters per forward pass. It is optimized for general-purpose text generation, including instruction following, logical reasoning, math, code, and tool usage. The model supports a native 262K context length and does not implement "thinking mode" (<think> blocks). Compared to its base variant, this version delivers significant gains in knowledge coverage, long-context reasoning, coding benchmarks, and alignment with open-ended tasks. It is particularly strong on multilingual understanding, math reasoning (e.g., AIME, HMMT), and alignment evaluations like Arena-Hard and WritingBench.
Input
Output
Context
262K
Max Output
—
Parameters
235B
Technical Specifications
Capabilities
Input Modalities
Output Modalities
Benchmarks
Pricing
per 1M tokens • updated regularlyPricing
Input
$0.071
Output
$0.100
Available on 1 Platform
| Platform | Input | Output |
|---|---|---|
OpenRouter | $0.071 | $0.100 |
Estimated GPU Requirements for Qwen3 235B A22B Instruct 2507
Estimates based on INT8 quantization. Actual requirements vary by framework and configuration.
Related Tools
Data sourced from official provider APIs and documentation
Last updated: Mar 17, 2026
Build Your AI Stack with Confidence
Explore models, compare pricing and benchmarks, and right-size your infrastructure — all in one place.