Qwen3 235B A22B 2507
Review Qwen3 235B A22B 2507 on key metrics including price, context length, throughput, and model features.
AuthorQwen
Context Length262.1k
Supports Tools
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. 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.
Activity
Last 14 daysPrompt
80KCompletion
3KTotal
83KStartup
Startup
Latency (p50)0.30s
Throughput (p50)147.9 tok/s
Pricing
Input$0.07/M tokens
Output$0.42/M tokens
Features
Input Modalitiestext
Output Modalitiestext
Supported EndpointsChat Completions
Vision
Supports Tools