Qwen3.5 35B A3B vs Seed 1.8
Wins 0 of 4 areas
—
Wins 3 of 4 areas
Coding · Agents · Following instructions
Seed 1.8 is the stronger all-rounder.
Scores updated · 27 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Seed 1.8 pulls ahead
- Completes tasks by operating a computer desktopOSWorld-Verified+7.4points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+6.7points ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+6.6points ahead
Where Qwen3.5 35B A3B pulls ahead
- Reasoning about charts from research papersCharXiv (RQ)+6.1points ahead
- Questions that truly need the image to answerMMStar+2points ahead
Every test, side by side
All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingSeed 1.8
- LiveCodeBench v6Seed 1.8 by 4.974.679.5+4.9
- SWE-bench VerifiedSeed 1.8 by 3.769.272.9+3.7
AgentsSeed 1.8
- OSWorld-VerifiedSeed 1.8 by 7.454.561.9+7.4
- BrowseCompSeed 1.8 by 6.66167.6+6.6
Images and chartsEven
- CharXiv (RQ)Qwen3.5 35B A3B by 6.177.571.4+6.1
- MMMUSeed 1.8 by 281.483.4+2
- MMStarQwen3.5 35B A3B by 281.979.9+2
- MathVistaSeed 1.8 by 1.586.287.7+1.5
- MMMU-Protie72.773.2tie
Following instructionsSeed 1.8
- Multi-ChallengeSeed 1.8 by 6.76066.7+6.7
Other results17 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 35B A3B by 23.58561.5+23.5
- BrowseComp-ZHSeed 1.8 by 11.869.581.3+11.8
- BFCLv4Qwen3.5 35B A3B by 10.167.357.2+10.1
- RefSpatial-BenchQwen3.5 35B A3B by 7.263.556.3+7.2
- SimpleVQASeed 1.8 by 7.158.365.4+7.1
- WideSearchSeed 1.8 by 6.757.163.8+6.7
- ERQAQwen3.5 35B A3B by 664.858.8+6
- Terminal-Bench 2.0Seed 1.8 by 4.740.545.2+4.7
- HallusionBenchQwen3.5 35B A3B by 467.963.9+4
- AI2DQwen3.5 35B A3B by 3.592.689.1+3.5
- ZeroBenchSeed 1.8 by 3811+3
- MATH-VisionQwen3.5 35B A3B by 2.683.981.3+2.6
- VideoMMMUSeed 1.8 by 2.380.482.7+2.3
- LVBenchSeed 1.8 by 1.671.473+1.6
- SuperGPQASeed 1.8 by 1.463.464.8+1.4
- MMVUtie72.373.1tie
- MMLU-Protie85.384.9tie
Questions people ask
Which is better, Qwen3.5 35B A3B or Seed 1.8?
Seed 1.8 wins three of the four areas where both have results: coding, agents and following instructions. Qwen3.5 35B A3B wins none. They are level on images and charts.
Which is better for coding?
Seed 1.8. It wins 2 of the 2 coding tests both models report; Qwen3.5 35B A3B wins none.
Which is cheaper?
Qwen3.5 35B A3B costs $0.25 per million input tokens and $2.00 per million output tokens; Seed 1.8 costs $0.25 and $2.00. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 27 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 17 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.