Qwen3.5 122B A10B vs Qwen3.5 397B A17B
Wins 1 of 7 areas
Agents
Wins 5 of 7 areas
Coding · Reasoning · Images and charts · Long documents · Following instructions
Qwen3.5 397B A17B is the stronger all-rounder.Qwen3.5 122B A10B is cheaper and better at agents.
Scores updated · 49 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 software08Qwen3.5 397B A17B8 of 8 tests
- ReasoningHard problems that need careful thinking03Qwen3.5 397B A17B3 of 3 tests
- Images and chartsUnderstanding pictures, charts and video03Qwen3.5 397B A17B3 of 3 tests
- Long documentsFinding answers in very long texts02Qwen3.5 397B A17B2 of 2 tests
- Following instructionsDoing exactly what it is asked02Qwen3.5 397B A17B2 of 2 tests
- AgentsCarrying out multi-step tasks on its own21Qwen3.5 122B A10B2 of 4 tests · 1 tie
- FactsGetting facts right instead of making them up11Even1 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 122B A10B costs 14% 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 397B A17B pulls ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+9.8points ahead
- Fixes real GitHub issues in many programming languagesSWE-bench Multilingual+7.3points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+6.4points ahead
Where Qwen3.5 122B A10B pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+1.9points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+1.8points ahead
- Real work tasks from 44 professionsGDPVal+1.1points ahead
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.
CodingQwen3.5 397B A17B
- Terminal-Bench HardQwen3.5 397B A17B by 9.831.140.9+9.8
- SWE-bench MultilingualQwen3.5 397B A17B by 7.36269.3+7.3
- SWE-bench ProQwen3.5 397B A17B by 644.950.9+6
- SciCodeQwen3.5 397B A17B by 5.139.744.8+5.1
- LiveCodeBench v6Qwen3.5 397B A17B by 4.778.983.6+4.7
- SWE-bench VerifiedQwen3.5 397B A17B by 4.47276.4+4.4
- LMArena · WebDevQwen3.5 397B A17B by 40 rating points13601400+40 rating
- Terminal-Bench 2.1Qwen3.5 397B A17B by 3.747.651.3+3.7
AgentsQwen3.5 122B A10B
- BrowseCompQwen3.5 397B A17B by 5.263.869+5.2
- τ-Bench V3 · BankingQwen3.5 122B A10B by 1.915.313.4+1.9
- GDPValQwen3.5 122B A10B by 1.115.714.6+1.1
- Terminal-Bench 4.0tie00tie
ReasoningQwen3.5 397B A17B
- Humanity's Last ExamQwen3.5 397B A17B by 3.825.229+3.8
- GPQA DiamondQwen3.5 397B A17B by 3.685.789.3+3.6
- CritPtQwen3.5 397B A17B by 1.10.61.7+1.1
FactsEven
- AA-Omniscience · AccuracyQwen3.5 397B A17B by 6.424.430.8+6.4
- AA-Omniscience · Non-hallucinationQwen3.5 122B A10B by 1.812.911.1+1.8
Images and chartsQwen3.5 397B A17B
- MMMU-ProQwen3.5 397B A17B by 2.37577.3+2.3
- LMArena · VisionQwen3.5 397B A17B by 18 rating points12451263+18 rating
- MMMUQwen3.5 397B A17B by 1.183.985+1.1
Long documentsQwen3.5 397B A17B
- LongBench v2Qwen3.5 397B A17B by 360.263.2+3
- AA-LCRQwen3.5 397B A17B by 176.377.3+1
Following instructionsQwen3.5 397B A17B
- Multi-ChallengeQwen3.5 397B A17B by 6.161.567.6+6.1
- IFBenchQwen3.5 397B A17B by 3.175.778.8+3.1
Other results25 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.
- VITA-BenchQwen3.5 397B A17B by 16.133.649.7+16.1
- WideSearchQwen3.5 397B A17B by 13.560.574+13.5
- AA-OmniscienceQwen3.5 397B A17B by 10.8-41.5-30.8+10.8
- DeepPlanningQwen3.5 397B A17B by 10.224.134.3+10.2
- GDPval-AA v2Qwen3.5 397B A17B by 101 rating points861962+101 rating
- t2-benchQwen3.5 397B A17B by 7.279.586.7+7.2
- SuperGPQAQwen3.5 397B A17B by 3.367.170.4+3.3
- Terminal-Bench 2.0Qwen3.5 397B A17B by 3.149.452.5+3.1
- AA IntelligenceQwen3.5 397B A17B by 2.815.618.4+2.8
- IncludeQwen3.5 397B A17B by 2.882.885.6+2.8
- Seal-0Qwen3.5 397B A17B by 2.844.146.9+2.8
- Artificial Analysis Coding IndexQwen3.5 397B A17B by 2.545.748.2+2.5
- MMLU-ProXQwen3.5 397B A17B by 2.582.284.7+2.5
- HMMT 2025Qwen3.5 397B A17B by 2.490.392.7+2.4
- τ²-Bench Telecom (AA run)Qwen3.5 397B A17B by 293.695.6+2
- MMMLUQwen3.5 397B A17B by 1.886.788.5+1.8
- C-EvalQwen3.5 397B A17B by 1.191.993+1.1
- MMLU-ProQwen3.5 397B A17B by 1.186.787.8+1.1
- AA Agentic IndexQwen3.5 397B A17B by 19.610.6+1
- MMLU-Reduxtie9494.9tie
- IFEvaltie93.492.6tie
- τ³-Bench Bankingtie14.213.4tie
- BFCLv4tie72.272.9tie
- WMT24++tie78.378.9tie
- BrowseComp-ZHtie69.970.3tie
Questions people ask
Which is better, Qwen3.5 122B A10B or Qwen3.5 397B A17B?
Qwen3.5 397B A17B wins five of the seven areas where both have results: coding, reasoning, images and charts, long documents and following instructions. Qwen3.5 122B A10B wins agents, and costs 14% less. They are level on facts.
Which is better for coding?
Qwen3.5 397B A17B. It wins 8 of the 8 coding tests both models report; Qwen3.5 122B A10B wins none.
Which is cheaper?
Qwen3.5 122B A10B costs $0.40 per million input tokens and $3.20 per million output tokens; Qwen3.5 397B A17B costs $0.60 and $3.60. That makes Qwen3.5 122B A10B about 14% 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 24 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 25 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.