Qwen3.5 122B A10B vs Qwen3.5 35B A3B
Wins 6 of 7 areas
Coding · Agents · Reasoning · Images and charts · Long documents · Following instructions
Wins 0 of 7 areas
—
Qwen3.5 122B A10B is the stronger all-rounder.Qwen3.5 35B A3B is cheaper.
Both rank among the ten best models we track in images and charts.
Scores updated · 75 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 software70Qwen3.5 122B A10B7 of 8 tests · 1 tie
- Images and chartsUnderstanding pictures, charts and video50Qwen3.5 122B A10B5 of 6 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own40Qwen3.5 122B A10B4 of 4 tests
- ReasoningHard problems that need careful thinking20Qwen3.5 122B A10B2 of 3 tests · 1 tie
- Long documentsFinding answers in very long texts20Qwen3.5 122B A10B2 of 2 tests
- Following instructionsDoing exactly what it is asked20Qwen3.5 122B A10B2 of 2 tests
- FactsGetting facts right instead of making them up11Even1 each · 1 tie
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.
Qwen3.5 35B A3B costs 38% 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 Qwen3.5 122B A10B pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+10.4points ahead
- Real work tasks from 44 professionsGDPVal+10.3points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+6.8points ahead
Where Qwen3.5 35B A3B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+1.7points ahead
Every test, side by side
All 75 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.5 122B A10B
- LMArena · WebDevQwen3.5 122B A10B by 110 rating points13601250+110 rating
- Terminal-Bench 2.1Qwen3.5 122B A10B by 6.847.640.8+6.8
- Terminal-Bench HardQwen3.5 122B A10B by 4.631.126.5+4.6
- LiveCodeBench v6Qwen3.5 122B A10B by 4.378.974.6+4.3
- SWE-bench VerifiedQwen3.5 122B A10B by 2.87269.2+2.8
- SciCodeQwen3.5 122B A10B by 239.737.7+2
- SWE-bench MultilingualQwen3.5 122B A10B by 1.76260.3+1.7
- SWE-bench Protie44.944.6tie
AgentsQwen3.5 122B A10B
- τ-Bench V3 · BankingQwen3.5 122B A10B by 10.415.34.9+10.4
- GDPValQwen3.5 122B A10B by 10.315.75.4+10.3
- OSWorld-VerifiedQwen3.5 122B A10B by 3.55854.5+3.5
- BrowseCompQwen3.5 122B A10B by 2.863.861+2.8
ReasoningQwen3.5 122B A10B
- Humanity's Last ExamQwen3.5 122B A10B by 4.225.221+4.2
- GPQA DiamondQwen3.5 122B A10B by 1.285.784.5+1.2
- CritPttie0.60.9tie
FactsEven
- AA-Omniscience · AccuracyQwen3.5 122B A10B by 4.324.420.1+4.3
- AA-Omniscience · Non-hallucinationQwen3.5 35B A3B by 1.712.914.6+1.7
- Vectara HHEM hallucination ratelower is bettertie11.210.5tie
Images and chartsQwen3.5 122B A10B
Long documentsQwen3.5 122B A10B
- AA-LCRQwen3.5 122B A10B by 4.376.372+4.3
- LongBench v2Qwen3.5 122B A10B by 1.260.259+1.2
Following instructionsQwen3.5 122B A10B
- IFBenchQwen3.5 122B A10B by 3.275.772.5+3.2
- Multi-ChallengeQwen3.5 122B A10B by 1.561.560+1.5
Other results47 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.
- Terminal-Bench 2.0Qwen3.5 122B A10B by 8.949.440.5+8.9
- Artificial Analysis Coding IndexQwen3.5 122B A10B by 8.745.737+8.7
- vectara_avg_summary_lengthQwen3.5 35B A3B by 8.586.494.9+8.5
- AA-OmniscienceQwen3.5 122B A10B by 6.6-41.5-48.1+6.6
- RefSpatial-BenchQwen3.5 122B A10B by 5.869.363.5+5.8
- BFCLv4Qwen3.5 122B A10B by 4.972.267.3+4.9
- τ²-Bench Telecom (AA run)Qwen3.5 122B A10B by 4.493.689.2+4.4
- AA IntelligenceQwen3.5 35B A3B by 3.715.619.3+3.7
- SuperGPQAQwen3.5 122B A10B by 3.767.163.4+3.7
- OJBenchQwen3.5 122B A10B by 3.539.536+3.5
- SimpleVQAQwen3.5 122B A10B by 3.461.758.3+3.4
- WideSearchQwen3.5 122B A10B by 3.460.557.1+3.4
- IncludeQwen3.5 122B A10B by 3.182.879.7+3.1
- LVBenchQwen3.5 122B A10B by 374.471.4+3
- ERQAQwen3.5 35B A3B by 2.86264.8+2.8
- Seal-0Qwen3.5 122B A10B by 2.744.141.4+2.7
- MMVUQwen3.5 122B A10B by 2.474.772.3+2.4
- MATH-VisionQwen3.5 122B A10B by 2.386.283.9+2.3
- AA Agentic IndexQwen3.5 35B A3B by 2.29.611.8+2.2
- WMT24++Qwen3.5 122B A10B by 278.376.3+2
- BabyVisionQwen3.5 122B A10B by 1.840.238.4+1.8
- MV-BenchQwen3.5 122B A10B by 1.876.674.8+1.8
- ScreenSpot-Pro (No tools)Qwen3.5 122B A10B by 1.870.468.6+1.8
- C-EvalQwen3.5 122B A10B by 1.791.990.2+1.7
- MLVUQwen3.5 122B A10B by 1.787.385.6+1.7
- t2-benchQwen3.5 35B A3B by 1.779.581.2+1.7
- VITA-BenchQwen3.5 122B A10B by 1.733.631.9+1.7
- VideoMMMUQwen3.5 122B A10B by 1.68280.4+1.6
- IFEvalQwen3.5 122B A10B by 1.593.491.9+1.5
- MMMLUQwen3.5 122B A10B by 1.586.785.2+1.5
- MMLU-ProQwen3.5 122B A10B by 1.486.785.3+1.4
- DeepPlanningQwen3.5 122B A10B by 1.324.122.8+1.3
- MMLU-ProXQwen3.5 122B A10B by 1.282.281+1.2
- CC-OCRQwen3.5 122B A10B by 1.181.880.7+1.1
- HMMT 2025Qwen3.5 122B A10B by 1.190.389.2+1.1
- RealWorldQAQwen3.5 122B A10B by 185.184.1+1
- ZeroBenchQwen3.5 122B A10B by 198+1
- DynaMathtie85.985tie
- EmbSpatial-Benchtie83.983.1tie
- AI2Dtie93.392.6tie
- MMLU-Reduxtie9493.3tie
- vectara_factual_consistencytie88.889.5tie
- MMlongBench Doctie5959.5tie
- OmniDocBench 1.5tie89.889.3tie
- BrowseComp-ZHtie69.969.5tie
- HallusionBenchtie67.667.9tie
- vectara_answer_ratetie99.899.8tie
Questions people ask
Which is better, Qwen3.5 122B A10B or Qwen3.5 35B A3B?
Qwen3.5 122B A10B wins six of the seven areas where both have results: coding, agents, reasoning, images and charts, long documents and following instructions. Qwen3.5 35B A3B wins none, but costs 38% less. They are level on facts.
Which is better for coding?
Qwen3.5 122B A10B. It wins 7 of the 8 coding tests both models report; Qwen3.5 35B A3B wins none, and 1 is a tie.
Which is cheaper?
Qwen3.5 122B A10B costs $0.40 per million input tokens and $3.20 per million output tokens; Qwen3.5 35B A3B costs $0.25 and $2.00. That makes Qwen3.5 35B A3B about 38% cheaper for the same work.
How do you compare the two?
We use the 75 benchmark tests both models have published scores on. The verdict counts the 28 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 47 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.