Nemotron 3.5 Super VL vs Qwen3.8 27B
Wins 1 of 6 areas
Images and charts
Wins 3 of 6 areas
Coding · Agents · Reasoning
Qwen3.8 27B wins more areas, narrowly.Nemotron 3.5 Super VL is cheaper and better at images and charts.
Scores updated · 43 tests both models report · How we compare
Where each one wins
Tests won in each of the six 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.
- CodingWriting and fixing software13Qwen3.8 27B3 of 4 tests
- AgentsCarrying out multi-step tasks on its own13Qwen3.8 27B3 of 4 tests
- ReasoningHard problems that need careful thinking12Qwen3.8 27B2 of 3 tests
- Images and chartsUnderstanding pictures, charts and video10Nemotron 3.5 Super VL1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
- Long documentsFinding answers in very long texts11Even1 each
Images and charts 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.
Nemotron 3.5 Super VL costs 81% 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 27B pulls ahead
- Abstract visual puzzles that people can solveARC-AGI-2+40.9points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+19.1points ahead
- Reasons across sets of long documentsAA-LCR+13points ahead
Where Nemotron 3.5 Super VL pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+17.3points ahead
- Questions about very long textsLongBench v2+9.8points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+7.7points ahead
Every test, side by side
All 43 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen3.8 27B
- Terminal-Bench 2.1Qwen3.8 27B by 19.160.779.8+19.1
- SWE-bench VerifiedQwen3.8 27B by 6.771.177.8+6.7
- SWE-bench MultilingualQwen3.8 27B by 6.167.773.8+6.1
- SciCodeNemotron 3.5 Super VL by 1.247.846.6+1.2
AgentsQwen3.8 27B
- BrowseCompNemotron 3.5 Super VL by 17.378.961.6+17.3
- GDPValQwen3.8 27B by 12.933.146+12.9
- MCP AtlasQwen3.8 27B by 9.667.777.3+9.6
- Terminal-Bench 4.0Qwen3.8 27B by 5.605.6+5.6
ReasoningQwen3.8 27B
- ARC-AGI-2Qwen3.8 27B by 40.91.542.4+40.9
- CritPtNemotron 3.5 Super VL by 5.210.65.4+5.2
- Humanity's Last ExamQwen3.8 27B by 3.230.733.9+3.2
FactsEven
- AA-Omniscience · AccuracyNemotron 3.5 Super VL by 7.723.315.6+7.7
- AA-Omniscience · Non-hallucinationQwen3.8 27B by 5.564.269.7+5.5
Images and chartsNemotron 3.5 Super VL
- OCRBenchv2Nemotron 3.5 Super VL by 3.967.463.5+3.9
Long documentsEven
- AA-LCRQwen3.8 27B by 136982+13
- LongBench v2Nemotron 3.5 Super VL by 9.862.152.3+9.8
Other results27 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.
- ARC-AGI-1Qwen3.8 27B by 64.822.787.5+64.8
- AutomationBench-AA v1.0.6Qwen3.8 27B by 33.414.848.2+33.4
- MRCR v2 8-needle 256K–512KQwen3.8 27B by 32.35.738+32.3
- SWEBench Pro v2 (Hard)Qwen3.8 27B by 29.744.173.8+29.7
- GDPval-AA 2.1Qwen3.8 27B by 265 rating points11621427+265 rating
- CritPt v2 FullNemotron 3.5 Super VL by 26.336.910.6+26.3
- BVBQwen3.8 27B by 18.652.170.7+18.6
- DeepSWE 1.1Qwen3.8 27B by 17.524.742.2+17.5
- OmniDocBench v1.6Qwen3.8 27B by 13.677.991.5+13.6
- GraphWalks BFS 256KNemotron 3.5 Super VL by 12.963.350.4+12.9
- Covert misbehavior (Internal) ↓Nemotron 3.5 Super VL by 10.243.433.2+10.2
- CritPt v2 HardNemotron 3.5 Super VL by 9.513.54+9.5
- GDP (Surge AI)Qwen3.8 27B by 8.4816.4+8.4
- AA IntelligenceQwen3.8 27B by 7.226.533.7+7.2
- HLE-DiamondQwen3.8 27B by 6.58.815.3+6.5
- AA-OmniscienceNemotron 3.5 Super VL by 5.8-4.2-10+5.8
- Terminal-Bench 4.0Qwen3.8 27B by 5.605.6+5.6
- Misbehavior (Internal) ↓Qwen3.8 27B by 4.952.857.7+4.9
- MMMU Pro Standard 10Qwen3.8 27B by 472.376.3+4
- JobBenchNemotron 3.5 Super VL by 3.136.533.4+3.1
- Chartography originalQwen3.8 27B by 2.59.712.2+2.5
- DeepSearchQA (F1)Qwen3.8 27B by 277.779.7+2
- ZeroBench (main) originalQwen3.8 27B by 11112+1
- StrongREJECT (none)tie99.499tie
- OCRBench v2 (Chinese)tie56.656.9tie
- ComplexConstraints v1.1tie30.130.1tie
- Terminal-Bench-Science 0.1tie00tie
Questions people ask
Which is better, Nemotron 3.5 Super VL or Qwen3.8 27B?
Qwen3.8 27B wins three of the six areas where both have results: coding, agents and reasoning. Nemotron 3.5 Super VL wins images and charts, and costs 81% less. They are level on facts and long documents.
Which is better for coding?
Qwen3.8 27B. It wins 3 of the 4 coding tests both models report; Nemotron 3.5 Super VL wins 1.
Which is cheaper?
Nemotron 3.5 Super VL costs $0.15 per million input tokens and $0.50 per million output tokens; Qwen3.8 27B costs $0.50 and $3.00. That makes Nemotron 3.5 Super VL about 81% cheaper for the same work.
How do you compare the two?
We use the 43 benchmark tests both models have published scores on. The verdict counts the 16 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 27 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.