Qwen3.8 27B vs Qwen3.8 Flash Next
Wins 1 of 8 areas
Long documents
Wins 4 of 8 areas
Coding · Reasoning · Images and charts · Following instructions
Qwen3.8 Flash Next wins more areas, narrowly.Qwen3.8 27B is better at long documents.
Both rank among the ten best models we track in following instructions.
Scores updated · 42 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 software15Qwen3.8 Flash Next5 of 8 tests · 2 ties
- ReasoningHard problems that need careful thinking04Qwen3.8 Flash Next4 of 4 tests
- Following instructionsDoing exactly what it is asked02Qwen3.8 Flash Next2 of 2 tests
- Images and chartsUnderstanding pictures, charts and video01Qwen3.8 Flash Next1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Qwen3.8 27B1 of 1 test
- AgentsCarrying out multi-step tasks on its own22Even2 each
- FactsGetting facts right instead of making them up11Even1 each
- MathCompetition and research-level math00Even0 each · 1 tie
Math and long documents 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 82% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Qwen3.8 Flash Next pulls ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+19.7points ahead
- Real work tasks from 44 professionsGDPVal+10.4points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+8.9points ahead
Where Qwen3.8 27B pulls ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+28.5points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+15points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+3.1points ahead
Every test, side by side
All 42 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.8 Flash Next
- SWE-bench MultilingualQwen3.8 Flash Next by 7.273.881+7.2
- Terminal-Bench 2.1Qwen3.8 Flash Next by 6.379.886.1+6.3
- SciCodeQwen3.8 Flash Next by 446.650.6+4
- LMArena · WebDevQwen3.8 Flash Next by 39 rating points15941633+39 rating
- LiveBench · CodingQwen3.8 27B by 3.175.772.5+3.1
- LiveCodeBench v6Qwen3.8 Flash Next by 1.690.391.9+1.6
- SWE-bench Protie61.762.5tie
- LiveBench · Agentic Codingtie61.461.6tie
AgentsEven
- OSWorld-VerifiedQwen3.8 27B by 28.584.355.8+28.5
- Terminal-Bench 4.0Qwen3.8 Flash Next by 19.75.625.3+19.7
- GDPValQwen3.8 Flash Next by 10.44656.4+10.4
- τ-Bench V3 · BankingQwen3.8 27B by 2.64845.4+2.6
ReasoningQwen3.8 Flash Next
- LiveBench · ReasoningQwen3.8 Flash Next by 7.48087.4+7.4
- CritPtQwen3.8 Flash Next by 5.75.411.1+5.7
- Humanity's Last ExamQwen3.8 Flash Next by 4.133.938+4.1
- GPQA DiamondQwen3.8 Flash Next by 1.890.592.3+1.8
FactsEven
- AA-Omniscience · Non-hallucinationQwen3.8 27B by 1569.754.7+15
- AA-Omniscience · AccuracyQwen3.8 Flash Next by 8.915.624.5+8.9
Images and chartsQwen3.8 Flash Next
- MMMU-ProQwen3.8 Flash Next by 3.576.379.8+3.5
- CharXiv (RQ)tie90.290.6tie
Following instructionsQwen3.8 Flash Next
- LiveBench · Instruction FollowingQwen3.8 Flash Next by 4.472.777.1+4.4
- IFBenchQwen3.8 Flash Next by 1.879.581.3+1.8
Other results18 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.
- BabyVisionQwen3.8 27B by 38.885.646.8+38.8
- JobBenchQwen3.8 Flash Next by 22.333.455.7+22.3
- DeepSWE 1.1Qwen3.8 Flash Next by 16.542.258.7+16.5
- AA Agentic IndexQwen3.8 Flash Next by 9.946.556.4+9.9
- Agents' Last ExamQwen3.8 Flash Next by 8.342.951.2+8.3
- ERQAQwen3.8 Flash Next by 6.865.572.3+6.8
- Toolathlon VerifiedQwen3.8 Flash Next by 6.467.173.5+6.4
- AA IntelligenceQwen3.8 Flash Next by 6.133.739.8+6.1
- NL2RepoQwen3.8 Flash Next by 5.842.348.1+5.8
- Artificial Analysis Coding IndexQwen3.8 Flash Next by 568.173.1+5
- OSWorld 2.0 (partial)Qwen3.8 Flash Next by 4.34852.3+4.3
- LVBenchQwen3.8 Flash Next by 4.272.476.6+4.2
- AndroidWorldQwen3.8 Flash Next by 2.681.984.5+2.6
- RealWorldQAQwen3.8 Flash Next by 2.685.988.5+2.6
- livebench_data_analysisQwen3.8 27B by 2.476.674.2+2.4
- MATH-VisionQwen3.8 Flash Next by 1.194.695.7+1.1
- AA-Omnisciencetie-10-9.7tie
- livebench_languagetie74.374.6tie
Questions people ask
Which is better, Qwen3.8 27B or Qwen3.8 Flash Next?
Qwen3.8 Flash Next wins four of the eight areas we test: coding, reasoning, images and charts and following instructions. Qwen3.8 27B wins long documents. They are level on agents, facts and math.
Which is better for coding?
Qwen3.8 Flash Next. It wins 5 of the 8 coding tests both models report; Qwen3.8 27B wins 1, and 2 are ties.
Which is cheaper?
Qwen3.8 27B costs $0.50 per million input tokens and $3.00 per million output tokens; Qwen3.8 Flash Next costs $0.15 and $0.47. That makes Qwen3.8 Flash Next about 82% cheaper for the same work.
How do you compare the two?
We use the 42 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 18 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.