Qwen3 235B A22B
Zero-eval
#1Arena Hard
#1BBH
#1CRUX-O
+5 more
by Alibaba
About
Qwen3 235B A22B is a language model developed by Alibaba. It achieves strong performance with an average score of 76.2% across 23 benchmarks. It excels particularly in Arena Hard (95.6%), GSM8k (94.4%), BBH (88.9%). The model shows particular specialization in math tasks with an average performance of 83.3%. It supports a 256K token context window for handling large documents. The model is available through 4 API providers. It's licensed for commercial use, making it suitable for enterprise applications. Released in 2025, it represents Alibaba's latest advancement in AI technology.
Pricing Range
Input (per 1M)$0.10 -$0.20
Output (per 1M)$0.10 -$0.80
Providers4
Timeline
AnnouncedApr 29, 2025
ReleasedApr 29, 2025
Specifications
Training Tokens36.0T
License & Family
License
Apache 2.0
Benchmark Performance Overview
Performance metrics and category breakdown
Overall Performance
23 benchmarks
Average Score
76.2%
Best Score
95.6%
High Performers (80%+)
10Performance Metrics
Max Context Window
256.0KAvg Throughput
38.0 tok/sAvg Latency
1msTop Categories
math
83.3%
roleplay
77.1%
code
76.6%
general
74.8%
Benchmark Performance
Top benchmark scores with normalized values (0-100%)
Ranking Across Benchmarks
Position relative to other models on each benchmark
Arena Hard
Rank #1 of 22
#1Qwen3 235B A22B
95.6%
#2Qwen3 32B
93.8%
#3DeepSeek-R1
92.3%
#4Qwen3 30B A3B
91.0%
GSM8k
Rank #16 of 46
#13Gemma 3 12B
94.4%
#14Nova Lite
94.5%
#15Qwen2.5 14B Instruct
94.8%
#16Qwen3 235B A22B
94.4%
#17Mistral Large 2
93.0%
#18Claude 3 Sonnet
92.3%
#19Nova Micro
92.3%
BBH
Rank #1 of 8
#1Qwen3 235B A22B
88.9%
#2Nova Pro
86.9%
#3Qwen2.5 32B Instruct
84.5%
#4DeepSeek-V2.5
84.3%
MMLU
Rank #12 of 78
#9DeepSeek-V3
88.5%
#10GPT-4o
88.7%
#11Kimi K2 Instruct
89.5%
#12Qwen3 235B A22B
87.8%
#13Kimi K2 Base
87.8%
#14GPT-4.1 mini
87.5%
#15Grok-2
87.5%
MMLU-Redux
Rank #6 of 13
#3DeepSeek-V3
89.1%
#4Kimi K2 Instruct
92.7%
#5DeepSeek-R1
92.9%
#6Qwen3 235B A22B
87.4%
#7Qwen2.5 72B Instruct
86.8%
#8Qwen2.5 32B Instruct
83.9%
#9Qwen2.5 14B Instruct
80.0%
All Benchmark Results for Qwen3 235B A22B
Complete list of benchmark scores with detailed information
Arena Hard Arena Hard benchmark | general | text | 0.96 | 95.6% | Self-reported |
GSM8k GSM8k benchmark | math | text | 0.94 | 94.4% | Self-reported |
BBH BBH benchmark | general | text | 0.89 | 88.9% | Self-reported |
MMLU MMLU benchmark | general | text | 0.88 | 87.8% | Self-reported |
MMLU-Redux MMLU-Redux benchmark | general | text | 0.87 | 87.4% | Self-reported |
MMMLU MMMLU benchmark | general | text | 0.87 | 86.7% | Self-reported |
AIME 2024 AIME 2024 benchmark | general | text | 0.86 | 85.7% | Self-reported |
MGSM MGSM benchmark | math | text | 0.84 | 83.5% | Self-reported |
AIME 2025 AIME 2025 benchmark | general | text | 0.81 | 81.5% | Self-reported |
MBPP MBPP benchmark | code | text | 81.40 | 81.4% | Self-reported |
Showing 1 to 10 of 23 benchmarks