Which is better, DeepSeek-V4-Flash or Nemotron 3 Ultra 550B A55B?

Refresh Models —Deals —Regulatory —Sources —

DeepSeek-V4-Flash vs Nemotron 3 Ultra 550B A55B

56 SHARED BENCHMARKS

Across 56 shared benchmarks, DeepSeek-V4-Flash scores higher on 44 and Nemotron 3 Ultra 550B A55B on 12. The widest gap is critpt, where DeepSeek-V4-Flash scores 16.6 against 3.1. DeepSeek-V4-Flash is the cheaper of the two on tracked API pricing ($0.44 against $0.68 per million input tokens).

DEEPSEEKVSNVIDIA56 SHARED44–12 HEAD-TO-HEAD

BenchmarkDeepSeek-V4-FlashMARGINNemotron 3 Ultra 550B A55B

AA Agentic Index61.327.5

AA Intelligence51.838.3

AA-LCR74.371

AA-Omniscience-14.3-0.4

aime_202695.894.2

Apex (Pass@1)3384.8

Apex-Shortlist (no tools)82.474.9

Apex-Shortlist (with tools)8284.8

Artificial Analysis Coding Index69.149.3

browsecomp73.244.4

BrowseComp (context management)73.263

critpt16.63.1

CritPt (no tools)10.63.1

gdpval52.946.7

GDPval-AA v211891164

Global-MMLU-Lite88.485.6

GPQA Diamond90.887

HLE38.637.4

HLE (with tools)45.137.4

hmmt_feb_202694.878.8

IFBench79.281.7

imo_answer_bench89.692.3

livebench_agentic_coding46.816.3

livebench_coding7570.7

livebench_data_analysis79.354.5

livebench_instruction_following65.573.5

livebench_language79.270.8

livebench_math86.888.7

livebench_reasoning86.674.7

LiveCodeBench v690.989

MCP Atlas6944.7

MMLU-Pro86.486.8

OmniScience Accuracy40.422.6

OmniScience Non-Hallucination15.670.3

PinchBench91.390

ProfBench (Search)5756

scicode49.944.6

SciCode (subtask)48.244.6

simplebench46.341.7

simpleqa_verified4532.4

strongreject97.498.7

SWE-bench Multilingual73.367.7

SWE-bench Verified7970.7

SWEBench Pro Public52.646.4

Tau 3 Banking22.913.8

TauBench V3 - Airline80.881.5

TauBench V3 - Average73.770.9

TauBench V3 - Banking39.414.2

TauBench V3 - Retail89.186.4

Terminal-Bench 2.182.756.4

Terminal-Bench Hard38.636.4

Toolathlon Verified70.334.3

Vals.ai Financial Agent 1.1 - with web search60.153.7

Vals.ai Financial Agent 1.1 - without web search58.460.1

τ²-Bench Telecom (AA run)95.683.3

τ³-Bench26.722.6

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing. Quoted rates are the price-setter row we currently track for each model — its direct or vendor-official listing where one exists (direct, direct), otherwise the lowest tracked offer.