Gemini 4 Argon
Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
Kimi K3
GLM-5.3
GPT-5.5
ERNIE 5.1
DeepSeek-V4.1-Flash-20260910
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 PreviewClaude Opus 4.6 vs Qwen3.8-Max
- Claude Opus 4.6 rates higher on LMArena’s text leaderboard: 1502 against 1484.
- Qwen3.8-Max is cheaper: $2 / $6 per million input / output tokens, against $5 / $25 for Claude Opus 4.6.
- On the 7 Vals AI benchmarks both have taken, Claude Opus 4.6 wins 2 and Qwen3.8-Max wins 5 (Claude Opus 4.6: MedCode, MMLU-Pro).
LatestClaude Opus 4.6 1502Qwen3.8-Max 1484
Side by side
Models (2/6)
- Claude Opus 4.6
- Qwen3.8-Max
| Benchmark | Claude Opus 4.6 | Qwen3.8-Max |
|---|---|---|
| The model | ||
| Lab | Anthropic | Alibaba |
| Released | 2026-02-04 | 2026-09-03 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 1M |
| Input, per 1M tokens | $5 | $2 |
| Output, per 1M tokens | $25 | $6 |
| LMArena, people’s votes | ||
| LMArena TextMillions of people compare two anonymous models side by side and vote for the better answer | 1502#6 of 351 | 1484#15 of 351 |
| LMArena WebDevPeople ask two anonymous models to build the same web app, try both and vote for the better one | 1546#45 of 124 | 1673#12 of 124 |
| LMArena VisionSide-by-side votes on prompts that include an image | 1316#6 of 146 | 1313#10 of 146 |
| LMArena CodingLMArena’s text votes on prompts that ask for code | 1534#3 of 346 | 1503#21 of 346 |
| LMArena MathLMArena’s text votes on math prompts | 1518#7 of 339 | 1501#15 of 339 |
| LMArena Creative writingLMArena’s text votes on stories, poems and other creative writing | 1505#4 of 349 | 1477#16 of 349 |
| LMArena Hard promptsLMArena’s text votes on the hardest prompts | 1524#5 of 351 | 1497#16 of 351 |
| LMArena Instruction followingLMArena’s text votes on prompts with specific instructions to follow | 1521#3 of 351 | 1481#18 of 351 |
| Vals AI, overall | ||
| Vals IndexVals AI’s headline score | — | 59.05%#11 of 27 |
| Vals AI, coding | ||
| Terminal-Bench 4.0Frontier-difficulty tasks an agent completes in a real terminal | — | 34.34%#18 of 55 |
| SWE-benchReal GitHub issues from open-source projects | 78.20%#30 of 85 | 85.60%#16 of 85 |
| Vibe Code BenchBuilding complete web apps from a plain-language brief, graded on whether the app works | 53.50%#57 of 115 | 64.70%#49 of 115 |
| Code MigrationRewriting working programs in a new language, graded on whether the new program still passes its tests | — | 57.19%#10 of 28 |
| LiveCodeBenchFresh competitive programming problems, collected after models’ training cut-offs to avoid contamination | 84.68%#38 of 143 | 87.85%#10 of 143 |
| Vals AI, finance | ||
| Finance AgentThe core tasks of a financial analyst, done as an agent with research tools | — | 58.83%#23 of 31 |
| Finance IndexVals AI’s composite of finance tasks | — | 61.66%#5 of 31 |
| Tax Agent BenchAgentic research into US corporate tax questions | — | 32.09%#22 of 76 |
| Vals AI, law | ||
| Legal IndexVals AI’s composite of legal tasks | — | 73.90%#5 of 28 |
| Legal Research BenchAgentic US legal research | — | 46.15%#17 of 85 |
| Vals AI, medicine | ||
| MedCodeAssigning the right medical billing codes from clinical notes | 48.24%#30 of 107 | 40.67%#62 of 107 |
| MedQAMedical licensing exam questions, measured on accuracy and on robustness to bias | 95.41%#12 of 92 | — |
| Vals AI, reasoning | ||
| GPQA DiamondGraduate-level questions in biology, physics and chemistry that experts answer well and search does not | 89.65%#29 of 138 | 93.69%#6 of 138 |
| MMLU-ProHarder, reasoning-focused multiple choice across 14 subjects, from law to engineering | 89.11%#15 of 138 | 88.60%#18 of 138 |
| AIMEProblems from the American Invitational Mathematics Examination, a qualifier for the US Math Olympiad | 95.63%#9 of 96 | — |
| MMMUCollege-level questions that need both an image and text | 83.87%#29 of 93 | 88.03%#12 of 93 |
| Terminal-Bench ScienceExpert-written scientific research workflows, run by an agent in a terminal | — | 12.86%#13 of 42 |
Add models in the compare tool.
Questions
Which is better, Claude Opus 4.6 or Qwen3.8-Max?
Claude Opus 4.6 rates higher on LMArena’s text leaderboard: 1502 against 1484. Qwen3.8-Max is cheaper: $2 / $6 per million input / output tokens, against $5 / $25 for Claude Opus 4.6. On the 7 Vals AI benchmarks both have taken, Claude Opus 4.6 wins 2 and Qwen3.8-Max wins 5 (Claude Opus 4.6: MedCode, MMLU-Pro).
Which is cheaper, Claude Opus 4.6 or Qwen3.8-Max?
Qwen3.8-Max is cheaper: $2 / $6 per million input / output tokens, against $5 / $25 for Claude Opus 4.6.