Claude Opus 4 vs Qwen3 235B A22B
Wins 4 of 5 areas
Coding · Reasoning · Long documents · Following instructions
Wins 1 of 5 areas
Facts
Claude Opus 4 is the stronger all-rounder.Qwen3 235B A22B is cheaper and better at facts.
Scores updated · 49 tests both models report · How we compare
Where each one wins
Tests won in each of the five areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software30Claude Opus 43 of 4 tests · 1 tie
- ReasoningHard problems that need careful thinking20Claude Opus 42 of 2 tests
- Long documentsFinding answers in very long texts20Claude Opus 42 of 2 tests
- Following instructionsDoing exactly what it is asked20Claude Opus 42 of 2 tests
- FactsGetting facts right instead of making them up01Qwen3 235B A22B1 of 1 test
Facts rests on a single test.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Qwen3 235B A22B costs 96% less for the same work.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Claude Opus 4 pulls ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+38.1points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+25points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+17.4points ahead
Where Qwen3 235B A22B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 49 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 4
- SWE-bench VerifiedClaude Opus 4 by 38.172.534.4+38.1
- Terminal-Bench HardClaude Opus 4 by 2531.16.1+25
- LiveCodeBench v6Claude Opus 4 by 10.447.437+10.4
- SciCodetie39.839.9tie
ReasoningClaude Opus 4
- GPQA DiamondClaude Opus 4 by 9.679.670+9.6
- Humanity's Last ExamClaude Opus 4 by 1.312.311+1.3
FactsQwen3 235B A22B
- Vectara HHEM hallucination ratelower is betterQwen3 235B A22B by 2.7129.3+2.7
Long documentsClaude Opus 4
- AA-LCRClaude Opus 4 by 69.369.30+69.3
- LongBench v2Claude Opus 4 by 5.555.650.1+5.5
Following instructionsClaude Opus 4
- Multi-ChallengeClaude Opus 4 by 17.458.641.2+17.4
- IFBenchClaude Opus 4 by 1553.738.7+15
Other results38 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- τ²-Bench Telecom (AA run)Claude Opus 4 by 49.473.424+49.4
- AIME 2024Qwen3 235B A22B by 37.548.285.7+37.5
- τ²-BenchClaude Opus 4 by 34.95722.1+34.9
- Tau2 airlineClaude Opus 4 by 33.56026.5+33.5
- TAU-bench (airline)Claude Opus 4 by 24.959.634.7+24.9
- τ²-Bench (Retail)Claude Opus 4 by 24.881.857+24.8
- MultiPL-EClaude Opus 4 by 23.789.665.9+23.7
- TAU-bench (retail)Claude Opus 4 by 22.881.458.6+22.8
- ZebraLogicClaude Opus 4 by 21.659.337.7+21.6
- OpenAI-MRCR (128k)Claude Opus 4 by 21.248.927.7+21.2
- livecodebench_hardQwen3 235B A22B by 20.933.154+20.9
- MMLU-ProClaude Opus 4 by 18.486.668.2+18.4
- Artificial Analysis Coding IndexClaude Opus 4 by 16.63417.4+16.6
- SuperGPQAClaude Opus 4 by 12.456.544.1+12.4
- AA IntelligenceClaude Opus 4 by 11.120.69.5+11.1
- SimpleQAClaude Opus 4 by 9.622.813.2+9.6
- LiveCodeBench (24/8~25/5)Qwen3 235B A22B by 9.356.665.9+9.3
- CNMO 2024Claude Opus 4 by 957.648.6+9
- livecodebench_mediumQwen3 235B A22B by 8.480.488.8+8.4
- OJBenchClaude Opus 4 by 8.319.611.3+8.3
- FullStackBenchClaude Opus 4 by 7.470.362.9+7.4
- MMLU-ReduxClaude Opus 4 by 6.894.287.4+6.8
- AIME 2025Qwen3 235B A22B by 675.581.5+6
- MMLUClaude Opus 4 by 5.992.987+5.9
- AceBenchClaude Opus 4 by 5.175.670.5+5.1
- IFEvalClaude Opus 4 by 4.287.483.2+4.2
- HMMT 2025Claude Opus 4 by 415.911.9+4
- vectara_answer_rateQwen3 235B A22B by 3.99194.9+3.9
- MATH-500 (EM)Claude Opus 4 by 3.294.491.2+3.2
- AutoLogiClaude Opus 4 by 2.886.183.3+2.8
- vectara_factual_consistencyQwen3 235B A22B by 2.78890.7+2.7
- HMMT Feb. 2025Qwen3 235B A22B by 2.56062.5+2.5
- LiveBenchQwen3 235B A22B by 2.574.677.1+2.5
- MMMLUClaude Opus 4 by 2.188.886.7+2.1
- PolyMath-enQwen3 235B A22B by 2.149.851.9+2.1
- vectara_avg_summary_lengthClaude Opus 4 by 17.6 rating points123.2105.6+17.6 rating
- LiveCodeBenchtie70.470.7tie
- livecodebench_easytie99.199.1tie
Questions people ask
Which is better, Claude Opus 4 or Qwen3 235B A22B?
Claude Opus 4 wins four of the five areas where both have results: coding, reasoning, long documents and following instructions. Qwen3 235B A22B wins facts, and costs 96% less.
Which is better for coding?
Claude Opus 4. It wins 3 of the 4 coding tests both models report; Qwen3 235B A22B wins none, and 1 is a tie.
Which is cheaper?
Claude Opus 4 costs $15.00 per million input tokens and $75.00 per million output tokens; Qwen3 235B A22B costs $0.70 and $2.80. That makes Qwen3 235B A22B about 96% cheaper for the same work.
How do you compare the two?
We use the 49 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 38 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.