Qwen3.6 27B vs Qwen3.8 27B
Wins 0 of 8 areas
—
Wins 7 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math · Long documents · Following instructions
Qwen3.8 27B is the stronger all-rounder.
Scores updated · 36 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 software07Qwen3.8 27B7 of 7 tests
- ReasoningHard problems that need careful thinking04Qwen3.8 27B4 of 4 tests
- AgentsCarrying out multi-step tasks on its own03Qwen3.8 27B3 of 3 tests
- Following instructionsDoing exactly what it is asked02Qwen3.8 27B2 of 2 tests
- Images and chartsUnderstanding pictures, charts and video12Qwen3.8 27B2 of 3 tests
- MathCompetition and research-level math01Qwen3.8 27B1 of 1 test
- Long documentsFinding answers in very long texts01Qwen3.8 27B1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
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 27B costs 17% 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 27B pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+31.3points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+22.1points ahead
- Real work tasks from 44 professionsGDPVal+21.8points ahead
Where Qwen3.6 27B pulls ahead
- Reads and reasons about text in imagesOCRBenchv2+4.8points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+4points ahead
Every test, side by side
All 36 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.8 27B
- LiveBench · Agentic CodingQwen3.8 27B by 22.139.361.4+22.1
- Terminal-Bench 2.1Qwen3.8 27B by 19.160.779.8+19.1
- SWE-bench ProQwen3.8 27B by 8.253.561.7+8.2
- LiveCodeBench v6Qwen3.8 27B by 6.483.990.3+6.4
- LiveBench · CodingQwen3.8 27B by 3.971.875.7+3.9
- SciCodeQwen3.8 27B by 3.842.846.6+3.8
- SWE-bench MultilingualQwen3.8 27B by 2.571.373.8+2.5
AgentsQwen3.8 27B
- τ-Bench V3 · BankingQwen3.8 27B by 31.316.748+31.3
- GDPValQwen3.8 27B by 21.824.446.2+21.8
- Terminal-Bench 4.0Qwen3.8 27B by 5.605.6+5.6
ReasoningQwen3.8 27B
- Humanity's Last ExamQwen3.8 27B by 10.823.133.9+10.8
- LiveBench · ReasoningQwen3.8 27B by 9.770.380+9.7
- GPQA DiamondQwen3.8 27B by 6.384.290.5+6.3
- CritPtQwen3.8 27B by 4.31.15.4+4.3
FactsEven
- AA-Omniscience · Non-hallucinationQwen3.8 27B by 1950.769.7+19
- AA-Omniscience · AccuracyQwen3.6 27B by 419.615.6+4
Images and chartsQwen3.8 27B
- CharXiv (RQ)Qwen3.8 27B by 11.878.490.2+11.8
- OCRBenchv2Qwen3.6 27B by 4.868.363.5+4.8
- MMMU-ProQwen3.8 27B by 1.774.676.3+1.7
Following instructionsQwen3.8 27B
- LiveBench · Instruction FollowingQwen3.8 27B by 19.553.272.7+19.5
- IFBenchQwen3.8 27B by 11.967.679.5+11.9
Other results13 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.
- DeepSWE 1.1Qwen3.8 27B by 28.913.342.2+28.9
- AA Agentic IndexQwen3.8 27B by 26.420.146.5+26.4
- Artificial Analysis Coding IndexQwen3.8 27B by 14.453.768.1+14.4
- AA IntelligenceQwen3.8 27B by 12.321.433.7+12.3
- AndroidWorldQwen3.8 27B by 11.670.381.9+11.6
- JobBenchQwen3.8 27B by 11.621.833.4+11.6
- livebench_languageQwen3.8 27B by 11.163.374.3+11.1
- AA-OmniscienceQwen3.8 27B by 10-20-10+10
- livebench_data_analysisQwen3.8 27B by 6.270.476.6+6.2
- NL2RepoQwen3.8 27B by 6.136.242.3+6.1
- ERQAQwen3.8 27B by 362.565.5+3
- OCRBench v2 (Chinese)Qwen3.6 27B by 1.858.756.9+1.8
- RealWorldQAQwen3.8 27B by 1.884.185.9+1.8
Questions people ask
Which is better, Qwen3.6 27B or Qwen3.8 27B?
Qwen3.8 27B wins seven of the eight areas we test: coding, agents, reasoning, images and charts, math, long documents and following instructions. Qwen3.6 27B wins none. They are level on facts.
Which is better for coding?
Qwen3.8 27B. It wins 7 of the 7 coding tests both models report; Qwen3.6 27B wins none.
Which is cheaper?
Qwen3.6 27B costs $0.60 per million input tokens and $3.60 per million output tokens; Qwen3.8 27B costs $0.50 and $3.00. That makes Qwen3.8 27B about 17% cheaper for the same work.
How do you compare the two?
We use the 36 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 13 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.