Back to Models
Qwen3 235b a22b 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
Parameters
235000000000 B
Context
262,144 tokens
Released
Invalid Date
Leaderboards
Average Score combining domain-specific Autobench scores; Higher is better
Performance vs. Industry Average
Context Window
Qwen3 235b a22b 2507 has a smaller context window than average (406k tokens), with a context window of 262k tokens.