Qwen3.8 27B vs Claude Sonnet 5.5
Wins 1 of 8 areas
Following instructions
Wins 5 of 8 areas
Coding · Agents · Reasoning · Images and charts · Math
Claude Sonnet 5.5 is the stronger all-rounder.Qwen3.8 27B is cheaper and better at following instructions.
Scores updated · 33 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.
- ReasoningHard problems that need careful thinking05Claude Sonnet 5.55 of 5 tests
- CodingWriting and fixing software15Claude Sonnet 5.55 of 6 tests
- AgentsCarrying out multi-step tasks on its own02Claude Sonnet 5.52 of 2 tests
- Images and chartsUnderstanding pictures, charts and video01Claude Sonnet 5.51 of 1 test
- MathCompetition and research-level math01Claude Sonnet 5.51 of 1 test
- Following instructionsDoing exactly what it is asked10Qwen3.8 27B1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Images and charts, math, 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 27B costs 71% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Sonnet 5.5 pulls ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+58points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+41.5points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+38.4points ahead
Where Qwen3.8 27B pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+16.7points ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+15.9points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+5.1points ahead
Every test, side by side
All 33 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 5.5
- SWE-bench ProClaude Sonnet 5.5 by 19.661.781.3+19.6
- LMArena · WebDevClaude Sonnet 5.5 by 182 rating points15901772+182 rating
- SWE-bench MultilingualClaude Sonnet 5.5 by 16.573.890.3+16.5
- LiveBench · CodingClaude Sonnet 5.5 by 15.775.791.4+15.7
- SciCodeClaude Sonnet 5.5 by 14.446.661+14.4
- LiveBench · Agentic CodingQwen3.8 27B by 5.161.456.3+5.1
AgentsClaude Sonnet 5.5
- Terminal-Bench 4.0Claude Sonnet 5.5 by 585.663.6+58
- GDPValClaude Sonnet 5.5 by 20.94666.9+20.9
ReasoningClaude Sonnet 5.5
- ARC-AGI-2Claude Sonnet 5.5 by 41.542.483.9+41.5
- CritPtClaude Sonnet 5.5 by 265.431.4+26
- Humanity's Last ExamClaude Sonnet 5.5 by 21.133.955+21.1
- SimpleBenchClaude Sonnet 5.5 by 15.760.275.9+15.7
- LiveBench · ReasoningClaude Sonnet 5.5 by 11.68091.6+11.6
FactsEven
- AA-Omniscience · AccuracyClaude Sonnet 5.5 by 38.415.654+38.4
- AA-Omniscience · Non-hallucinationQwen3.8 27B by 16.769.753+16.7
Images and chartsClaude Sonnet 5.5
- LMArena · VisionClaude Sonnet 5.5 by 15 rating points12711286+15 rating
Following instructionsQwen3.8 27B
- LiveBench · Instruction FollowingQwen3.8 27B by 15.972.756.8+15.9
Other results14 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.
- Terminal-Bench 4.0Claude Sonnet 5.5 by 655.670.6+65
- CritPt v2 HardClaude Sonnet 5.5 by 55459+55
- Terminal-Bench-Science 0.1Claude Sonnet 5.5 by 53.3053.3+53.3
- AA-OmniscienceClaude Sonnet 5.5 by 42.3-1032.3+42.3
- GDPval-AA 2.1Claude Sonnet 5.5 by 417 rating points14271844+417 rating
- HLE-DiamondClaude Sonnet 5.5 by 3815.353.3+38
- DeepSWE 1.1Claude Sonnet 5.5 by 28.842.271+28.8
- AA IntelligenceClaude Sonnet 5.5 by 22.333.756+22.3
- livebench_data_analysisQwen3.8 27B by 17.176.659.5+17.1
- SWE-bench MultimodalClaude Sonnet 5.5 by 15.738.654.3+15.7
- Toolathlon VerifiedClaude Sonnet 5.5 by 10.767.177.8+10.7
- GDP (Surge AI)Claude Sonnet 5.5 by 9.416.425.8+9.4
- ARC-AGI-1Claude Sonnet 5.5 by 987.596.5+9
- livebench_languageClaude Sonnet 5.5 by 3.674.378+3.6
Questions people ask
Which is better, Qwen3.8 27B or Claude Sonnet 5.5?
Claude Sonnet 5.5 wins five of the eight areas we test: coding, agents, reasoning, images and charts and math. Qwen3.8 27B wins following instructions, and costs 71% less. They are level on facts and long documents.
Which is better for coding?
Claude Sonnet 5.5. It wins 5 of the 6 coding tests both models report; Qwen3.8 27B wins 1.
Which is cheaper?
Qwen3.8 27B costs $0.50 per million input tokens and $3.00 per million output tokens; Claude Sonnet 5.5 costs $2.00 and $10.00. That makes Qwen3.8 27B about 71% cheaper for the same work.
How do you compare the two?
We use the 33 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 14 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.