DeepSeek-V4-Flash vs NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16
Wins 3 of 4 areas
Coding · Reasoning · Following instructions
Wins 1 of 4 areas
Agents
DeepSeek-V4-Flash is the stronger all-rounder.NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 is better at agents.
Scores updated · 22 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.
- CodingWriting and fixing software40DeepSeek-V4-Flash4 of 5 tests · 1 tie
- ReasoningHard problems that need careful thinking20DeepSeek-V4-Flash2 of 2 tests
- Following instructionsDoing exactly what it is asked10DeepSeek-V4-Flash1 of 1 test
- AgentsCarrying out multi-step tasks on its own01NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF161 of 1 test
Agents 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.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where DeepSeek-V4-Flash pulls ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+18points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+8.3points ahead
- Fixes real GitHub issues in Python projectsSWE-bench Verified+7.9points ahead
Where NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+4.8points ahead
Every test, side by side
All 22 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingDeepSeek-V4-Flash
- Terminal-Bench 2.1DeepSeek-V4-Flash by 1878.760.7+18
- SWE-bench VerifiedDeepSeek-V4-Flash by 7.97971.1+7.9
- SWE-bench MultilingualDeepSeek-V4-Flash by 5.673.367.7+5.6
- LiveCodeBench v6DeepSeek-V4-Flash by 2.490.688.2+2.4
- SWE-bench Protie52.651.8tie
AgentsNVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16
- BrowseCompNVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 by 4.873.278+4.8
ReasoningDeepSeek-V4-Flash
- CritPtDeepSeek-V4-Flash by 5.716.610.9+5.7
- GPQA DiamondDeepSeek-V4-Flash by 1.390.889.5+1.3
Following instructionsDeepSeek-V4-Flash
- IFBenchDeepSeek-V4-Flash by 8.379.270.9+8.3
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.
- tau2-bench telecomDeepSeek-V4-Flash by 2598.273.2+25
- DeepSWE [offline]DeepSeek-V4-Flash by 17.642.324.6+17.6
- MMLU-ProXDeepSeek-V4-Flash by 8.685.576.9+8.6
- tau2-bench airlineDeepSeek-V4-Flash by 783.576.5+7
- BrowseComp [offline]DeepSeek-V4-Flash by 58378+5
- tau2-bench retailDeepSeek-V4-Flash by 4.673.869.2+4.6
- HLE, text-onlyDeepSeek-V4-Flash by 435.831.8+4
- τ³-Bench BankingDeepSeek-V4-Flash by 422.918.9+4
- LMArena v3 proxyNVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 by 2.949.852.7+2.9
- Terminal Bench 4 [offline]DeepSeek-V4-Flash by 2.53.51+2.5
- GDPval-AA v2NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 by 22 rating points11891211+22 rating
- SciCode, subtaskDeepSeek-V4-Flash by 2.249.347.1+2.2
- WMT24++tie85.584.8tie
Questions people ask
Which is better, DeepSeek-V4-Flash or NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16?
DeepSeek-V4-Flash wins three of the four areas where both have results: coding, reasoning and following instructions. NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 wins agents.
Which is better for coding?
DeepSeek-V4-Flash. It wins 4 of the 5 coding tests both models report; NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 wins none, and 1 is a tie.
How do you compare the two?
We use the 22 benchmark tests both models have published scores on. The verdict counts the 9 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.