Qwen3.5 122B A10B vs Qwen3.6 35B A3B
Wins 5 of 7 areas
Agents · Reasoning · Images and charts · Long documents · Following instructions
Wins 1 of 7 areas
Coding
Qwen3.5 122B A10B is the stronger all-rounder.Qwen3.6 35B A3B is cheaper and better at coding.
Scores updated · 52 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.
- Images and chartsUnderstanding pictures, charts and video20Qwen3.5 122B A10B2 of 4 tests · 2 ties
- ReasoningHard problems that need careful thinking20Qwen3.5 122B A10B2 of 3 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own21Qwen3.5 122B A10B2 of 4 tests · 1 tie
- Long documentsFinding answers in very long texts10Qwen3.5 122B A10B1 of 1 test
- Following instructionsDoing exactly what it is asked10Qwen3.5 122B A10B1 of 1 test
- CodingWriting and fixing software25Qwen3.6 35B A3B5 of 7 tests
- FactsGetting facts right instead of making them up11Even1 each
Long documents and following instructions 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 27% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3.5 122B A10B pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+15.1points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+11.3points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+6points ahead
Where Qwen3.6 35B A3B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+36.6points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+5.2points ahead
- Long, multi-file coding tasks in real codebasesSWE-bench Pro+4.6points ahead
Every test, side by side
All 52 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.6 35B A3B
- SWE-bench MultilingualQwen3.6 35B A3B by 5.26267.2+5.2
- SWE-bench ProQwen3.6 35B A3B by 4.644.949.5+4.6
- Terminal-Bench HardQwen3.6 35B A3B by 3.731.134.8+3.7
- SciCodeQwen3.5 122B A10B by 3.139.736.6+3.1
- Terminal-Bench 2.1Qwen3.5 122B A10B by 2.747.644.9+2.7
- LiveCodeBench v6Qwen3.6 35B A3B by 1.578.980.4+1.5
- SWE-bench VerifiedQwen3.6 35B A3B by 1.47273.4+1.4
AgentsQwen3.5 122B A10B
- BrowseCompQwen3.5 122B A10B by 15.163.848.7+15.1
- τ-Bench V3 · BankingQwen3.5 122B A10B by 615.39.3+6
- GDPValQwen3.6 35B A3B by 3.915.719.6+3.9
- Terminal-Bench 4.0tie00tie
ReasoningQwen3.5 122B A10B
- Humanity's Last ExamQwen3.5 122B A10B by 325.222.2+3
- GPQA DiamondQwen3.5 122B A10B by 1.685.784.1+1.6
- CritPttie0.60.3tie
FactsEven
- AA-Omniscience · Non-hallucinationQwen3.6 35B A3B by 36.612.949.5+36.6
- AA-Omniscience · AccuracyQwen3.5 122B A10B by 5.624.418.8+5.6
Images and chartsQwen3.5 122B A10B
- MMMUQwen3.5 122B A10B by 2.283.981.7+2.2
- MathVistaQwen3.5 122B A10B by 187.486.4+1
- CharXiv (RQ)tie77.278tie
- MMMU-Protie7575tie
Long documentsQwen3.5 122B A10B
- AA-LCRQwen3.5 122B A10B by 4.676.371.7+4.6
Following instructionsQwen3.5 122B A10B
- IFBenchQwen3.5 122B A10B by 11.375.764.4+11.3
Other results30 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.
- AA-OmniscienceQwen3.6 35B A3B by 19.3-41.5-22.2+19.3
- GDPval-AA v2Qwen3.6 35B A3B by 154 rating points8611015+154 rating
- BFCLv4Qwen3.5 122B A10B by 7.272.265+7.2
- τ³-Bench BankingQwen3.5 122B A10B by 6.614.27.6+6.6
- AA Agentic IndexQwen3.6 35B A3B by 5.49.615+5.4
- RefSpatial-BenchQwen3.5 122B A10B by 569.364.3+5
- Artificial Analysis Coding IndexQwen3.5 122B A10B by 3.845.741.9+3.8
- LVBenchQwen3.5 122B A10B by 374.471.4+3
- SimpleVQAQwen3.5 122B A10B by 2.861.758.9+2.8
- AA IntelligenceQwen3.6 35B A3B by 2.615.618.2+2.6
- SuperGPQAQwen3.5 122B A10B by 2.467.164.7+2.4
- HallusionBenchQwen3.6 35B A3B by 2.267.669.8+2.2
- Terminal-Bench 2.0Qwen3.6 35B A3B by 2.149.451.5+2.1
- MV-BenchQwen3.5 122B A10B by 276.674.6+2
- VITA-BenchQwen3.6 35B A3B by 233.635.6+2
- C-EvalQwen3.5 122B A10B by 1.991.990+1.9
- DeepPlanningQwen3.6 35B A3B by 1.824.125.9+1.8
- VideoMMMUQwen3.6 35B A3B by 1.78283.7+1.7
- τ²-Bench Telecom (AA run)Qwen3.6 35B A3B by 1.793.695.3+1.7
- MMLU-ProQwen3.5 122B A10B by 1.586.785.2+1.5
- HMMT 2025Qwen3.5 122B A10B by 1.290.389.1+1.2
- MLVUQwen3.5 122B A10B by 1.187.386.2+1.1
- MMLU-Reduxtie9493.3tie
- AI2Dtie93.392.7tie
- EmbSpatial-Benchtie83.984.3tie
- WideSearchtie60.560.1tie
- IFEvaltie93.493.1tie
- RealWorldQAtie85.185.3tie
- CC-OCRtie81.881.9tie
- OmniDocBench 1.5tie89.889.9tie
Questions people ask
Which is better, Qwen3.5 122B A10B or Qwen3.6 35B A3B?
Qwen3.5 122B A10B wins five of the seven areas where both have results: agents, reasoning, images and charts, long documents and following instructions. Qwen3.6 35B A3B wins coding, and costs 27% less. They are level on facts.
Which is better for coding?
Qwen3.6 35B A3B. It wins 5 of the 7 coding tests both models report; Qwen3.5 122B A10B wins 2.
Which is cheaper?
Qwen3.5 122B A10B costs $0.40 per million input tokens and $3.20 per million output tokens; Qwen3.6 35B A3B costs $0.38 and $2.25. That makes Qwen3.6 35B A3B about 27% cheaper for the same work.
How do you compare the two?
We use the 52 benchmark tests both models have published scores on. The verdict counts the 22 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 30 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.