qwen3-235b-a22b-2507
by qwenQwen3-235B-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. Optimized for general-purpose text generation, including instruction following, logical reasoning, math, code, and tool usage. Supports a native 262K context length and delivers significant gains in knowledge coverage, long-context reasoning, and coding benchmarks.
Pricing
Pay-as-you-go rates for this model. More details can be found here.
Input Tokens (1M)
$0.07
Output Tokens (1M)
$0.42
Capabilities
Input Modalities
Text
Output Modalities
Text
Rate Limits
Requests per minute (RPM) and per day (RPD) by tier. More about tiers here
| Tier | RPM | RPD |
|---|---|---|
| Free | — | — |
| Tier 1 | 10 | — |
| Tier 2 | 15 | — |
| Tier 3 | 25 | — |
| Tier 4 | 50 | — |
Usage Analytics
Token usage across the last 30 active days