Qwen3.5 122B A10B vs Qwen3.8 Flash Next
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Long documents · Following instructions
Qwen3.8 Flash Next is the stronger all-rounder.
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 software07Qwen3.8 Flash Next7 of 7 tests
- AgentsCarrying out multi-step tasks on its own14Qwen3.8 Flash Next4 of 5 tests
- ReasoningHard problems that need careful thinking03Qwen3.8 Flash Next3 of 3 tests
- Images and chartsUnderstanding pictures, charts and video02Qwen3.8 Flash Next2 of 2 tests
- FactsGetting facts right instead of making them up01Qwen3.8 Flash Next1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts01Qwen3.8 Flash Next1 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen3.8 Flash Next1 of 1 test
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.
Qwen3.8 Flash Next costs 83% 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 Qwen3.8 Flash Next pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+41.8points ahead
- Real work tasks from 44 professionsGDPVal+40.7points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+38.5points ahead
Where Qwen3.5 122B A10B pulls ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+2.2points 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.
CodingQwen3.8 Flash Next
- Terminal-Bench 2.1Qwen3.8 Flash Next by 38.547.686.1+38.5
- LMArena · WebDevQwen3.8 Flash Next by 273 rating points13601633+273 rating
- SWE-bench MultilingualQwen3.8 Flash Next by 196281+19
- SWE-bench ProQwen3.8 Flash Next by 17.644.962.5+17.6
- LiveCodeBench v6Qwen3.8 Flash Next by 1378.991.9+13
- SciCodeQwen3.8 Flash Next by 10.939.750.6+10.9
- SWE-bench VerifiedQwen3.8 Flash Next by 6.77278.7+6.7
AgentsQwen3.8 Flash Next
- GDPValQwen3.8 Flash Next by 40.715.756.4+40.7
- τ-Bench V3 · BankingQwen3.8 Flash Next by 30.115.345.4+30.1
- Terminal-Bench 4.0Qwen3.8 Flash Next by 25.3025.3+25.3
- BrowseCompQwen3.8 Flash Next by 25.263.889+25.2
- OSWorld-VerifiedQwen3.5 122B A10B by 2.25855.8+2.2
ReasoningQwen3.8 Flash Next
- Humanity's Last ExamQwen3.8 Flash Next by 12.825.238+12.8
- CritPtQwen3.8 Flash Next by 10.50.611.1+10.5
- GPQA DiamondQwen3.8 Flash Next by 6.685.792.3+6.6
FactsQwen3.8 Flash Next
- AA-Omniscience · Non-hallucinationQwen3.8 Flash Next by 41.812.954.7+41.8
- AA-Omniscience · Accuracytie24.424.5tie
Images and chartsQwen3.8 Flash Next
- CharXiv (RQ)Qwen3.8 Flash Next by 13.477.290.6+13.4
- MMMU-ProQwen3.8 Flash Next by 4.87579.8+4.8
Long documentsQwen3.8 Flash Next
- AA-LCRQwen3.8 Flash Next by 3.476.379.7+3.4
Following instructionsQwen3.8 Flash Next
- IFBenchQwen3.8 Flash Next by 5.675.781.3+5.6
Other results28 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 v2Qwen3.8 Flash Next by 700 rating points8611561+700 rating
- DeepSWE [offline]Qwen3.8 Flash Next by 55.20.455.6+55.2
- AA Agentic IndexQwen3.8 Flash Next by 46.89.656.4+46.8
- WebVoyager [offline]Qwen3.8 Flash Next by 35.554.990.4+35.5
- AA-OmniscienceQwen3.8 Flash Next by 31.8-41.5-9.7+31.8
- BrowseComp [offline]Qwen3.8 Flash Next by 315889+31
- τ³-Bench BankingQwen3.8 Flash Next by 30.314.244.5+30.3
- Artificial Analysis Coding IndexQwen3.8 Flash Next by 27.445.773.1+27.4
- AA IntelligenceQwen3.8 Flash Next by 24.215.639.8+24.2
- Terminal Bench 4 [offline]Qwen3.8 Flash Next by 20.7020.7+20.7
- tau2-bench airlineQwen3.8 Flash Next by 17.363.580.8+17.3
- tau2-bench retailQwen3.8 Flash Next by 16.255.471.6+16.2
- HLE, text-onlyQwen3.8 Flash Next by 15.424.239.6+15.4
- HLE, full setQwen3.8 Flash Next by 14.424.438.8+14.4
- Video-MME v2Qwen3.8 Flash Next by 13.13245.1+13.1
- LMArena v3 proxyQwen3.5 122B A10B by 11.84735.2+11.8
- ERQAQwen3.8 Flash Next by 10.36272.3+10.3
- MATH-VisionQwen3.8 Flash Next by 9.586.295.7+9.5
- MMlongBench DocQwen3.8 Flash Next by 8.95967.9+8.9
- WMT24++Qwen3.8 Flash Next by 7.978.386.2+7.9
- SciCode, subtaskQwen3.8 Flash Next by 6.839.846.6+6.8
- BabyVisionQwen3.8 Flash Next by 6.640.246.8+6.6
- RealWorldQAQwen3.8 Flash Next by 3.485.188.5+3.4
- tau2-bench telecomQwen3.8 Flash Next by 3.287.290.4+3.2
- LVBenchQwen3.8 Flash Next by 2.274.476.6+2.2
- OCRBench v2, ZHQwen3.8 Flash Next by 167.368.3+1
- MMLU-ProXtie82.283tie
- OCRBench v2_entie68.367.9tie
Questions people ask
Which is better, Qwen3.5 122B A10B or Qwen3.8 Flash Next?
Qwen3.8 Flash Next wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, long documents and following instructions. Qwen3.5 122B A10B wins none.
Which is better for coding?
Qwen3.8 Flash Next. It wins 7 of the 7 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; Qwen3.8 Flash Next costs $0.15 and $0.47. That makes Qwen3.8 Flash Next about 83% 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 21 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 28 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.