Model
Model explorer

Qwen3-235B-A22B-Instruct-2507

OPEN
Alibaba · Qwen3 family · released Jul 21, 2025

July refresh drops hybrid mode for a dedicated non-thinking Instruct model with native 256K context.

ReasoningCodingVisionFunction callingTool useAgentic
1591.0
Elo · rank #103
Parameters
235B
Active params
22B (MoE)
Context
256K tokens
Architecture
MoE, 128 experts / 8 active
License
Apache 2.0
Languages
119+
API price (in/out)
$0.23 / $0.92
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
Aider PolyglotCoding57.3%#20
best: Claude Opus 4.5 · 89.4%
AIMEMath70.3%#104
best: GPT-5.2 · 100.0%
Arena-Hard v2Human preference79.2%#3
best: Qwen3-Max · 86.1%
Arena EloHuman preference1423#15
best: Claude Fable 5 · 1505
BFCL v3Agents70.9%#5
best: Hunyuan-A13B · 78.3%
GPQA DiamondReasoning77.5%#92
best: GPT-5.6 · 94.6%
IFEvalReasoning88.7%#32
best: Gemma 4 26B A4B · 98.5%
LiveCodeBenchCoding51.8%#116
best: DeepSeek-V4-Pro (Think Max) · 93.5%
MMLU-ProKnowledge83.0%#39
best: Claude Fable 5 · 91.5%
MMLU-ReduxKnowledge93.1%#14
best: Qwen3.7-Max · 95.0%
SimpleQAKnowledge54.3%#3
best: GPT-4.5 · 62.5%
SuperGPQAReasoning62.6%#13
best: Qwen3.7-Max · 73.6%
tau-benchAgents71.3%#5
best: Claude Opus 4.1 · 82.4%
τ²-Bench TelecomAgents32.5%#35
best: Claude Opus 4.6 · 99.3%
Run it locally
VRAM @ Q4
142 GB
VRAM @ FP16
470 GB
Fits on (Q4)
M3 Ultra 512GBB200 192GB
Too large to run on a single RTX 4090 even at Q4; requires multi-GPU or CPU offload.
Quantizations
GGUF · FP8 · AWQ
Qwen3 family
Elo progression across releases
Qwen3-14B (Non-Thinking)Apr 20251094.6Qwen3-14B (Thinking)Apr 20251396.5Qwen3-235B-A22B (Non-Thinking)Apr 20251235.2Qwen3-235B-A22B (Thinking)Apr 20251536.2Qwen3-30B-A3B (Non-Thinking)Apr 20251101.1Qwen3-30B-A3B (Thinking)Apr 20251483.2Qwen3-32B (Non-Thinking)Apr 20251121.8Qwen3-32B (Thinking)Apr 20251509.2Qwen3-8B (Non-Thinking)Apr 2025922.1Qwen3-8B (Thinking)Apr 20251323.0Qwen3-235B-A22B-Instruct-2507Jul 20251591.0Qwen3-235B-A22B-Thinking-2507Jul 20251842.4
Resources
HFQwen/Qwen3-235B-A22B-Instruct-2507GHQwenLM/Qwen3DOCSqwen.readthedocs.ioAPIAvailable via Alibaba API
API price $0.23/$0.92 · each benchmark row carries its own source badge (see methodology)