Qwen 2.5 Max is the upgraded version of Qwen Max, beating GPT-4o, Deepseek V3 and Claude 3.5 Sonnet in benchmarks.
Added Oct 1, 2024
Context Window
32.0K
Max Output
8.2K
Input Price (Auto)
$1.60/1M
Output Price (Auto)
$6.39/1M
Cache Read (Auto)
$0.80/1M
Benchmarks
Benchmarks
Performance metrics and benchmarks
Sourced from LMArena.
Arena Score
1374.2
Overall Rank
#170 / 389
Votes
32,628
Confidence Interval
1370.1 - 1378.3
Category Scores
Coding
#186 / 384
5,101 votes
1402.9
Math
#178 / 374
3,306 votes
1363.1
Longer Query
#163 / 367
4,549 votes
1384.7
Creative Writing
#153 / 387
5,011 votes
1353.5
Instruction Following
#177 / 389
10,981 votes
1356.8
Hard Prompts
#175 / 389
9,687 votes
1385.3
Additional Categories22
French
#135 / 271
328 votes
1409.1
Japanese
#141 / 256
693 votes
1309.8
Spanish
#149 / 272
249 votes
1375.5
German
#154 / 293
760 votes
1353.9
Korean
#155 / 264
465 votes
1316.2
Polish
#157 / 215
862 votes
1358.2
Industry Writing And Literature And Language
#160 / 388
8,209 votes
1362.4
Industry Legal And Government
#164 / 361
1,871 votes
1386.6
Non English
#164 / 389
13,924 votes
1361.5
Russian
#164 / 352
2,808 votes
1366.0
Exclude Ties
#167 / 389
21,983 votes
1349.6
Industry Entertainment And Sports And Media
#167 / 387
5,997 votes
1340.0
Chinese
#168 / 360
1,951 votes
1396.6
Industry Life And Physical And Social Science
#168 / 387
5,703 votes
1389.7
Industry Medicine And Healthcare
#168 / 357
1,593 votes
1393.8
Multi Turn
#169 / 387
4,879 votes
1372.9
Industry Business And Management And Financial Operations
#172 / 382
3,685 votes
1368.9
Industry Mathematical
#176 / 368
2,950 votes
1367.3
English
#179 / 389
18,704 votes
1382.1
Expert
#182 / 339
1,689 votes
1367.5
Industry Software And It Services
#185 / 389
8,641 votes
1398.0
Hard Prompts English
#187 / 388
6,018 votes
1387.6
Published 2026-08-12 · Matched as qwen2.5-max
LMArena DatasetProviders
Auto routing is available for this model. Explicit provider selection is not available.
Loading provider options…
Related text models
Compare Qwen 2.5 Max with similar models from the same provider or model family.
Qwen3 Max
qwen/qwen3-maxQwen3 Max. The latest Qwen 3 model (5 september 2025). Higher accuracy in coding and science, better instruction following, and optimized for tool calling.
Qwen: QvQ Max
qvq-maxQvQ Max is the top model of the Qwen series. QvQ Max is capable of thinking and reasoning, can achieve significantly enhanced performance especially on hard problems.
Qwen 3.6 35B A3B Uncensored
qwen/qwen3.6-35b-a3b-uncensoredQwen 3.6 35B A3B Uncensored is an NVFP4 open-weight mixture-of-experts model LoRA-tuned for fewer refusals across chat, coding, tool use, and multimodal tasks.
Qwen 3.8 27B Uncensored
qwen/qwen3.8-27b-uncensoredQwen 3.8 27B Uncensored is an NVFP4 open-weight multimodal model LoRA-tuned for fewer refusals across chat, coding, tool use, and long-context work.
Qwen3.8 Max
qwen3.8-maxQwen3.8 Max is Qwen's 2.4T-parameter flagship model for coding, knowledge work, full-stack development, data analysis, and long-running agent workflows in non-thinking mode. It supports text, image, video, PDF input, tool calling, structured output, and a near-million-token context window.
Qwen3.8 Max Thinking
qwen3.8-max:thinkingQwen3.8 Max Thinking enables generation-time reasoning for deeper coding, knowledge work, data analysis, and long-running agent workflows. It supports text, image, video, PDF input, tool calling, structured output, and a near-million-token context window.