Claude Opus 5 vs Qwen3.6 35B A3B
Wins 6 of 7 areas
Coding · Agents · Reasoning · Images and charts · Math · Long documents
Wins 0 of 7 areas
—
Claude Opus 5 is the stronger all-rounder.Qwen3.6 35B A3B is cheaper.
Scores updated · 27 tests both models report · How we compare
Where each one wins
Tests won in each of the seven 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 software50Claude Opus 55 of 5 tests
- AgentsCarrying out multi-step tasks on its own50Claude Opus 55 of 5 tests
- ReasoningHard problems that need careful thinking30Claude Opus 53 of 3 tests
- Images and chartsUnderstanding pictures, charts and video20Claude Opus 52 of 2 tests
- MathCompetition and research-level math10Claude Opus 51 of 1 test
- Long documentsFinding answers in very long texts10Claude Opus 51 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Math and long documents rest on a single test each.
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.
Qwen3.6 35B A3B costs 91% 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 5 pulls ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+65.2points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+44.2points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+42.1points ahead
Where Qwen3.6 35B A3B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+10.3points ahead
Every test, side by side
All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Opus 5
- Terminal-Bench 2.1Claude Opus 5 by 44.289.144.9+44.2
- SWE-bench ProClaude Opus 5 by 29.779.249.5+29.7
- SWE-bench VerifiedClaude Opus 5 by 22.69673.4+22.6
- SWE-bench MultilingualClaude Opus 5 by 22.389.567.2+22.3
- SciCodeClaude Opus 5 by 19.856.436.6+19.8
AgentsClaude Opus 5
- Terminal-Bench 4.0Claude Opus 5 by 49490+49
- BrowseCompClaude Opus 5 by 42.190.848.7+42.1
- GDPValClaude Opus 5 by 41.561.119.6+41.5
- τ-Bench V3 · BankingClaude Opus 5 by 32.842.19.3+32.8
- MCP AtlasClaude Opus 5 by 2385.862.8+23
ReasoningClaude Opus 5
- Humanity's Last ExamClaude Opus 5 by 32.754.922.2+32.7
- CritPtClaude Opus 5 by 28.829.10.3+28.8
- GPQA DiamondClaude Opus 5 by 9.193.284.1+9.1
FactsEven
- AA-Omniscience · AccuracyClaude Opus 5 by 42.160.918.8+42.1
- AA-Omniscience · Non-hallucinationQwen3.6 35B A3B by 10.339.249.5+10.3
Images and chartsClaude Opus 5
- MMMU-ProClaude Opus 5 by 9.784.775+9.7
- CharXiv (RQ)Claude Opus 5 by 5.783.778+5.7
Other results8 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.
- GDPval-AA v2 EloClaude Opus 5 by 1278 rating points1824546+1278 rating
- GDPval-AA v2Claude Opus 5 by 809 rating points18241015+809 rating
- AA-OmniscienceClaude Opus 5 by 59.337.1-22.2+59.3
- AA Agentic IndexClaude Opus 5 by 41.256.215+41.2
- Artificial Analysis Coding IndexClaude Opus 5 by 36.17841.9+36.1
- AA IntelligenceClaude Opus 5 by 32.650.818.2+32.6
- LoCoMoClaude Opus 5 by 19.569.550+19.5
- LVBenchClaude Opus 5 by 475.471.4+4
Questions people ask
Which is better, Claude Opus 5 or Qwen3.6 35B A3B?
Claude Opus 5 wins six of the seven areas where both have results: coding, agents, reasoning, images and charts, math and long documents. Qwen3.6 35B A3B wins none, but costs 91% less. They are level on facts.
Which is better for coding?
Claude Opus 5. It wins 5 of the 5 coding tests both models report; Qwen3.6 35B A3B wins none.
Which is cheaper?
Claude Opus 5 costs $5.00 per million input tokens and $25.00 per million output tokens; Qwen3.6 35B A3B costs $0.38 and $2.25. That makes Qwen3.6 35B A3B about 91% cheaper for the same work.
How do you compare the two?
We use the 27 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 8 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.