Qwen3.5 27B vs Seed 1.8
Wins 2 of 4 areas
Coding · Images and charts
Wins 2 of 4 areas
Agents · Following instructions
The two are evenly matched.Qwen3.5 27B is better at images and charts and coding; Seed 1.8 at agents and following instructions.
Scores updated · 28 tests both models report · How we compare
Where each one wins
Tests won in each of the four 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.
Following instructions rests on a single test.
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.
Seed 1.8 costs 17% 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 27B pulls ahead
- Reasoning about charts from research papersCharXiv (RQ)+8.1points ahead
- Harder college exam questions with imagesMMMU-Pro+1.8points ahead
- Recent programming contest problemsLiveCodeBench v6+1.2points ahead
Where Seed 1.8 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+6.6points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+5.9points ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+5.7points ahead
Every test, side by side
All 28 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.5 27B
- LiveCodeBench v6Qwen3.5 27B by 1.280.779.5+1.2
- SWE-bench Verifiedtie72.472.9tie
AgentsSeed 1.8
- BrowseCompSeed 1.8 by 6.66167.6+6.6
- OSWorld-VerifiedSeed 1.8 by 5.756.261.9+5.7
Images and chartsQwen3.5 27B
- CharXiv (RQ)Qwen3.5 27B by 8.179.571.4+8.1
- MMMU-ProQwen3.5 27B by 1.87573.2+1.8
- MMMUSeed 1.8 by 1.182.383.4+1.1
- MMStarQwen3.5 27B by 1.18179.9+1.1
- MathVistatie87.887.7tie
Following instructionsSeed 1.8
- Multi-ChallengeSeed 1.8 by 5.960.866.7+5.9
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.
- DynaMathQwen3.5 27B by 26.287.761.5+26.2
- BrowseComp-ZHSeed 1.8 by 19.262.181.3+19.2
- RefSpatial-BenchQwen3.5 27B by 11.467.756.3+11.4
- BFCLv4Qwen3.5 27B by 11.368.557.2+11.3
- SimpleVQASeed 1.8 by 9.45665.4+9.4
- HallusionBenchQwen3.5 27B by 6.17063.9+6.1
- OSWorld-Verified (Overall)Seed 1.8 by 5.756.261.9+5.7
- MATH-VisionQwen3.5 27B by 4.78681.3+4.7
- AI2DQwen3.5 27B by 3.892.989.1+3.8
- Terminal-Bench 2.0Seed 1.8 by 3.641.645.2+3.6
- WideSearchSeed 1.8 by 2.761.163.8+2.7
- ERQAQwen3.5 27B by 1.760.558.8+1.7
- MMLU-ProQwen3.5 27B by 1.286.184.9+1.2
- ZeroBenchSeed 1.8 by 11011+1
- SuperGPQAtie65.664.8tie
- LVBenchtie73.673tie
- VideoMMMUtie82.382.7tie
- MMVUtie73.373.1tie
Questions people ask
Which is better, Qwen3.5 27B or Seed 1.8?
Qwen3.5 27B and Seed 1.8 each win two of the four areas where both have results. Qwen3.5 27B wins coding and images and charts; Seed 1.8 wins agents and following instructions. Seed 1.8 costs 17% less.
Which is better for coding?
Qwen3.5 27B. It wins 1 of the 2 coding tests both models report; Seed 1.8 wins none, and 1 is a tie.
Which is cheaper?
Qwen3.5 27B costs $0.30 per million input tokens and $2.40 per million output tokens; Seed 1.8 costs $0.25 and $2.00. That makes Seed 1.8 about 17% cheaper for the same work.
How do you compare the two?
We use the 28 benchmark tests both models have published scores on. The verdict counts the 10 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.