Qwen3.5 35B A3B vs Qwen3.5 4B
Wins 7 of 8 areas
Coding · Reasoning · Facts · Images and charts · Math · Long documents · Following instructions
Wins 0 of 8 areas
—
Qwen3.5 35B A3B is the stronger all-rounder.Qwen3.5 4B is cheaper.
Scores updated · 47 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. 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 software50Qwen3.5 35B A3B5 of 5 tests
- Images and chartsUnderstanding pictures, charts and video50Qwen3.5 35B A3B5 of 5 tests
- ReasoningHard problems that need careful thinking20Qwen3.5 35B A3B2 of 3 tests · 1 tie
- FactsGetting facts right instead of making them up20Qwen3.5 35B A3B2 of 2 tests
- MathCompetition and research-level math20Qwen3.5 35B A3B2 of 2 tests
- Long documentsFinding answers in very long texts20Qwen3.5 35B A3B2 of 2 tests
- Following instructionsDoing exactly what it is asked20Qwen3.5 35B A3B2 of 2 tests
- AgentsCarrying out multi-step tasks on its own11Even1 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.
Qwen3.5 4B costs 92% 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 35B A3B pulls ahead
Where Qwen3.5 4B pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+1.9points ahead
Every test, side by side
All 47 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.5 35B A3B
- SciCodeQwen3.5 35B A3B by 21.637.716.1+21.6
- LiveCodeBench v6Qwen3.5 35B A3B by 18.874.655.8+18.8
- Terminal-Bench 2.1Qwen3.5 35B A3B by 1540.825.8+15
- SWE-bench VerifiedQwen3.5 35B A3B by 9.769.259.5+9.7
- Terminal-Bench HardQwen3.5 35B A3B by 8.326.518.2+8.3
AgentsEven
- MCP AtlasQwen3.5 35B A3B by 2162.441.4+21
- τ-Bench V3 · BankingQwen3.5 4B by 1.94.96.8+1.9
ReasoningQwen3.5 35B A3B
- Humanity's Last ExamQwen3.5 35B A3B by 11.1219.9+11.1
- GPQA DiamondQwen3.5 35B A3B by 7.484.577.1+7.4
- CritPttie0.90tie
FactsQwen3.5 35B A3B
- AA-Omniscience · AccuracyQwen3.5 35B A3B by 520.115.1+5
- AA-Omniscience · Non-hallucinationQwen3.5 35B A3B by 1.214.613.4+1.2
Images and chartsQwen3.5 35B A3B
MathQwen3.5 35B A3B
- HMMT Feb. 2026Qwen3.5 35B A3B by 8.981.872.9+8.9
- AIME 2026Qwen3.5 35B A3B by 3.693.389.7+3.6
Long documentsQwen3.5 35B A3B
- AA-LCRQwen3.5 35B A3B by 97263+9
- LongBench v2Qwen3.5 35B A3B by 95950+9
Following instructionsQwen3.5 35B A3B
- IFBenchQwen3.5 35B A3B by 20.572.552+20.5
- Multi-ChallengeQwen3.5 35B A3B by 116049+11
Other results24 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 Agentic IndexQwen3.5 4B by 20.711.832.5+20.7
- ERQAQwen3.5 35B A3B by 18.564.846.3+18.5
- SimpleVQAQwen3.5 35B A3B by 17.658.340.7+17.6
- RealWorldQAQwen3.5 35B A3B by 1784.167.1+17
- BFCLv4Qwen3.5 35B A3B by 16.767.350.6+16.7
- HallusionBenchQwen3.5 35B A3B by 16.267.951.7+16.2
- Artificial Analysis Coding IndexQwen3.5 35B A3B by 14.43722.6+14.4
- HMMT 2025Qwen3.5 35B A3B by 12.489.276.8+12.4
- SuperGPQAQwen3.5 35B A3B by 10.563.452.9+10.5
- AA-OmniscienceQwen3.5 35B A3B by 10.2-48.1-58.4+10.2
- VITA-BenchQwen3.5 35B A3B by 9.931.922+9.9
- WMT24++Qwen3.5 35B A3B by 9.776.366.6+9.7
- MMLU-ProXQwen3.5 35B A3B by 9.58171.5+9.5
- MMMLUQwen3.5 35B A3B by 9.185.276.1+9.1
- IncludeQwen3.5 35B A3B by 8.779.771+8.7
- AA IntelligenceQwen3.5 35B A3B by 6.219.313.1+6.2
- MMLU-ProQwen3.5 35B A3B by 6.285.379.1+6.2
- DeepPlanningQwen3.5 35B A3B by 5.222.817.6+5.2
- C-EvalQwen3.5 35B A3B by 5.190.285.1+5.1
- MMLU-ReduxQwen3.5 35B A3B by 4.593.388.8+4.5
- Claw-Eval AvgQwen3.5 35B A3B by 3.165.462.3+3.1
- τ²-Bench Telecom (AA run)Qwen3.5 4B by 2.989.292.1+2.9
- IFEvalQwen3.5 35B A3B by 2.191.989.8+2.1
- t2-benchQwen3.5 35B A3B by 1.381.279.9+1.3
Questions people ask
Which is better, Qwen3.5 35B A3B or Qwen3.5 4B?
Qwen3.5 35B A3B wins seven of the eight areas we test: coding, reasoning, facts, images and charts, math, long documents and following instructions. Qwen3.5 4B wins none, but costs 92% less. They are level on agents.
Which is better for coding?
Qwen3.5 35B A3B. It wins 5 of the 5 coding tests both models report; Qwen3.5 4B wins none.
Which is cheaper?
Qwen3.5 35B A3B costs $0.25 per million input tokens and $2.00 per million output tokens; Qwen3.5 4B costs $0.03 and $0.15. That makes Qwen3.5 4B about 92% cheaper for the same work.
How do you compare the two?
We use the 47 benchmark tests both models have published scores on. The verdict counts the 23 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 24 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.