Model Cadence
Plus $4/mo

Qwen3 235B-A22B

Qwen · Released Apr 28, 2025 · Open weights

Large open Qwen MoE for multilingual reasoning, coding, and tool use

Strongest at
Coding
Top 32% of the past year's models · one benchmark so far
Price per 1M tokens
$0.7 / $2.8
input / output · Alibaba
Context window
131.1K
tokens in · 16.4K out
Open in the timeline

Capability profile

Scored against 90 models released in the past year. 50 is the average model; further out is better.

This modelMedian of the past year
5090Median of the past yearCoding: 59 · Top 32%Reasoning: 36 · Better than 10%Science: 49 · Better than 7%CodingReasoningScience
Coding59Top 32% · 1 test
Reasoning36Better than 10% · 1 test
Science49Better than 7%

No results yet: Writing, Reading & documents, Math, Visual reasoning, Computer use

Where these scores come from

Each capability combines the published benchmarks listed here, rescaled so 50 is the reference average and 15 points is one standard deviation. A hollow dot rests on a single benchmark.

Price against capability

The shaded corner holds anything better and cheaper. Prices blend input and output 3 to 1.

4 models from the past year score higher on Coding for less: DeepSeek V4.1 Flash, DeepSeek V3.2, Claude Haiku 5.5 and 1 more.

$0.1$0.3$1$3$10020406080Claude Haiku 5.5: 60, $0.2 per 1MMistral Large 4: 56, $2.06 per 1MGemini 4 Argon: 65, $4 per 1MGPT-6.1 Sol: 64, $4 per 1MClaude Sonnet 5.5: 71, $4 per 1MGPT-6 Sol: 62, $4 per 1MGPT-6 Luna: 55, $0.2 per 1MClaude Opus 5.5: 79, $8 per 1MGrok 4.7: 59, $3 per 1MDeepSeek V4.1 Flash: 72, $0.27 per 1MGPT-6 Astra: 73, $20 per 1MMuse Spark 1.3: 69, $2 per 1MGemini 3.8 Flash: 57, $1.5 per 1MClaude Fable 5.1: 85, $20 per 1MGLM-5.3-Flash: 54, $0.24 per 1MDeepSeek V4 Flash Vision Exp: 55, $0.32 per 1MQwen3.8 27B: 60, $1.12 per 1MGLM-5.3: 60, $2.15 per 1MGemini 3.7 Flash: 59, $1.5 per 1MDeepSeek V4 Pro 0813: 49, $0.99 per 1MGrok 4.6: 60, $3 per 1MMuse Spark 1.2: 53, $2 per 1MQwen3.8 Max: 55, $3 per 1MDeepSeek V4 Flash 0731: 38, $0.067 per 1MClaude Opus 5: 69, $10 per 1MGemini 3.6 Flash: 44, $1.5 per 1MGemini 3.5 Flash Lite: 41, $0.85 per 1MKimi K3: 63, $6 per 1MGPT-5.6 Terra: 58, $4.5 per 1MGPT-5.6 Sol: 68, $8 per 1MGPT-5.6 Luna: 56, $0.45 per 1MGrok 4.5: 50, $3 per 1MClaude Sonnet 5: 60, $4 per 1MGLM-5.2: 48, $2.15 per 1MKimi K2.7 Code: 36, $1.71 per 1MClaude Fable 5: 77, $20 per 1MQwen3.7 Plus: 25, $0.7 per 1MClaude Opus 4.8: 62, $10 per 1MQwen3.7 Max: 48, $3.75 per 1MGemini 3.5 Flash: 47, $3.38 per 1MDeepSeek V4 Pro: 41, $0.26 per 1MDeepSeek V4 Flash: 23, $0.11 per 1MGPT-5.5: 58, $11.2 per 1MQwen3.6 27B: 29, $1.35 per 1MKimi K2.6: 52, $1.71 per 1MGrok 4.3: 1, $1.56 per 1MGrok Build 0.1: 29, $1.25 per 1MClaude Opus 4.7: 60, $10 per 1MMuse Spark 1.1: 52, $2 per 1MGLM-5.1: 55, $2.15 per 1MQwen3.6 Plus: 39, $1.12 per 1MGPT-5.4 nano: 36, $0.46 per 1MGPT-5.4 mini: 37, $1.69 per 1MGPT-5.4: 52, $5.62 per 1MGemini 3.1 Pro Preview: 38, $4.5 per 1MClaude Sonnet 4.6: 35, $6 per 1MGLM-5: 52, $1.55 per 1MGPT-5.3 Codex: 56, $4.81 per 1MClaude Opus 4.6: 54, $10 per 1MKimi K2.5: 54, $1 per 1MGemini 3 Flash Preview: 49, $1.12 per 1MGPT-5.2 Codex: 55, $4.81 per 1MGPT-5.2: 57, $4.81 per 1MDeepSeek V3.2: 68, $0.32 per 1MClaude Opus 4.5: 59, $10 per 1MGPT-5.1: 46, $3.44 per 1MQwen3 235B-A22B: 59, $1.22 per 1MQwen3 235B-A22BPrice per 1M tokens →↑ Coding$0.1$1$10020406080Claude Haiku 5.5: 60, $0.2 per 1MMistral Large 4: 56, $2.06 per 1MGemini 4 Argon: 65, $4 per 1MGPT-6.1 Sol: 64, $4 per 1MClaude Sonnet 5.5: 71, $4 per 1MGPT-6 Sol: 62, $4 per 1MGPT-6 Luna: 55, $0.2 per 1MClaude Opus 5.5: 79, $8 per 1MGrok 4.7: 59, $3 per 1MDeepSeek V4.1 Flash: 72, $0.27 per 1MGPT-6 Astra: 73, $20 per 1MMuse Spark 1.3: 69, $2 per 1MGemini 3.8 Flash: 57, $1.5 per 1MClaude Fable 5.1: 85, $20 per 1MGLM-5.3-Flash: 54, $0.24 per 1MDeepSeek V4 Flash Vision Exp: 55, $0.32 per 1MQwen3.8 27B: 60, $1.12 per 1MGLM-5.3: 60, $2.15 per 1MGemini 3.7 Flash: 59, $1.5 per 1MDeepSeek V4 Pro 0813: 49, $0.99 per 1MGrok 4.6: 60, $3 per 1MMuse Spark 1.2: 53, $2 per 1MQwen3.8 Max: 55, $3 per 1MDeepSeek V4 Flash 0731: 38, $0.067 per 1MClaude Opus 5: 69, $10 per 1MGemini 3.6 Flash: 44, $1.5 per 1MGemini 3.5 Flash Lite: 41, $0.85 per 1MKimi K3: 63, $6 per 1MGPT-5.6 Terra: 58, $4.5 per 1MGPT-5.6 Sol: 68, $8 per 1MGPT-5.6 Luna: 56, $0.45 per 1MGrok 4.5: 50, $3 per 1MClaude Sonnet 5: 60, $4 per 1MGLM-5.2: 48, $2.15 per 1MKimi K2.7 Code: 36, $1.71 per 1MClaude Fable 5: 77, $20 per 1MQwen3.7 Plus: 25, $0.7 per 1MClaude Opus 4.8: 62, $10 per 1MQwen3.7 Max: 48, $3.75 per 1MGemini 3.5 Flash: 47, $3.38 per 1MDeepSeek V4 Pro: 41, $0.26 per 1MDeepSeek V4 Flash: 23, $0.11 per 1MGPT-5.5: 58, $11.2 per 1MQwen3.6 27B: 29, $1.35 per 1MKimi K2.6: 52, $1.71 per 1MGrok 4.3: 1, $1.56 per 1MGrok Build 0.1: 29, $1.25 per 1MClaude Opus 4.7: 60, $10 per 1MMuse Spark 1.1: 52, $2 per 1MGLM-5.1: 55, $2.15 per 1MQwen3.6 Plus: 39, $1.12 per 1MGPT-5.4 nano: 36, $0.46 per 1MGPT-5.4 mini: 37, $1.69 per 1MGPT-5.4: 52, $5.62 per 1MGemini 3.1 Pro Preview: 38, $4.5 per 1MClaude Sonnet 4.6: 35, $6 per 1MGLM-5: 52, $1.55 per 1MGPT-5.3 Codex: 56, $4.81 per 1MClaude Opus 4.6: 54, $10 per 1MKimi K2.5: 54, $1 per 1MGemini 3 Flash Preview: 49, $1.12 per 1MGPT-5.2 Codex: 55, $4.81 per 1MGPT-5.2: 57, $4.81 per 1MDeepSeek V3.2: 68, $0.32 per 1MClaude Opus 4.5: 59, $10 per 1MGPT-5.1: 46, $3.44 per 1MQwen3 235B-A22B: 59, $1.22 per 1MQwen3 235B-A22BPrice per 1M tokens →↑ Coding

Alibaba $0.7 / $2.8

The Qwen3.8 2.4T A95B line

7 releases in the Qwen3.8 2.4T A95B line.

Qwen3 235B-A22BApr 28, 20250dQwen3 30B A3BApr 28, 2025293dQwen3.5 397B-A17BFeb 15, 20268dQwen3.5 122B-A10BFeb 23, 20260dQwen3.5 35B-A3BFeb 23, 202651dQwen3.6 35B-A3BApr 15, 2026119dQwen3.8 2.4T A95BAug 12, 2026

A newer release is out: Qwen3.8 2.4T A95B.

ECI STANDING#128 of 270
SCORE139.15134.62–140.53 CI
FRONTIER LAGLaunched 4.3 months behindScores about what o1 did in December 2024.
SCORES LIKE THESE MODELSQwen3.5-9BAlibaba · Feb 2026139.32GPT-5 nanoOpenAI · Aug 2025139.21DeepSeek-R1DeepSeek · Jan 2025138.82
Qwen3-235B-A22B-Instruct (Jul 2025)Alibaba · Jul 2025138.78
Full Epoch Capabilities Index ranking ↗

Epoch benchmark boards

Best result on each board it has run. The tick marks the leader.

Fiction.LiveBench
68%#20 of 58
MATH level 5
69%#37 of 98
DTBench
76%#97 of 179
SimpleBench
31%#68 of 89

Qwen's past year

48 release days in the past year, about one every 8 days. The next is expected around Oct 30, 2026. Forecasts ↗

NovDecJanFebMarAprMayJunJulAugSepOct

Specifications

RELEASE DATEApr 28, 2025
WEIGHTSOpen weights
CONTEXT WINDOW131.1K
MAX OUTPUT16.4K
REASONINGNot listed
TOOL CALLINGNot listed
INPUTStext
OUTPUTStext

Models.dev record ↗

Specifications and prices from Models.dev; availability and limits may differ by provider. Benchmarks from Epoch AI (CC BY 4.0), LiveBench and the other sources named above.