Enter opens the highlighted result. Esc closes.

DeepSeek V4 Pro 0423 vs Nemotron 3 Ultra

DeepSeek V4 Pro 0423 leads Nemotron 3 Ultra on coding, 38.7 to 30.0 on the Vibe Coding Index, at about the same output price ($2.60 vs $2.40 per million). Updated October 22, 2026; 62 benchmarks have a result for both.

Replace DeepSeek V4 Pro 0423 Replace Nemotron 3 Ultra
DeepSeek V4 Pro 0423DeepSeek
Coding
38.7
Overall
50.0
Output $/M
$2.60
51
Nemotron 3 UltraNVIDIA
Coding
30.0
Overall
53.1
Output $/M
$2.40
10
vs boards won of 62 shared, 1 tied

DeepSeek V4 Pro 0423 or Nemotron 3 Ultra: which to pick

Pick DeepSeek V4 Pro 0423 for coding: it scores 38.7 on the Vibe Coding Index against 30.0 for Nemotron 3 Ultra.

Pick Nemotron 3 Ultra for output tokens 1.1x cheaper ($2.40 vs $2.60 per million) and work measured by Terminal-Bench 2.1 (Vals AI) or Mystery game puzzles, where it leads.

Model ADeepSeek V4 Pro 0423DeepSeek38.7/1005 independent benchmarks$0.57 per solved taskModel BNemotron 3 UltraNVIDIA30.0/1003 independent benchmarksNo cost per solved task
Shared results
62
DeepSeek V4 Pro 0423 only
27
Nemotron 3 Ultra only
9
Like-for-like categories
8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

At a glance

Coding index
DeepSeek V4 Pro 0423 38.7 from 5 benchmarks, Nemotron 3 Ultra 30.0 from 3 benchmarks.
Widest category
Web design, 10.3 points apart on average over 8 shared benchmarks.
Widest single gap
Vibe Code Bench (Vals AI): DeepSeek V4 Pro 0423 49.9% against Nemotron 3 Ultra 7.6%, 42.3 points apart on a 0 to 100 scale.
Output price
DeepSeek V4 Pro 0423 $2.60 against Nemotron 3 Ultra $2.40 per million output tokens, 1.1x apart.
Context window
DeepSeek V4 Pro 0423 takes 1.05M tokens, Nemotron 3 Ultra 262K.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
CategoryDeepSeek V4 Pro 0423BarsNemotron 3 UltraTies
Coding911 tie
Intelligence20
Agentic work40
Reasoning42
Math22
Knowledge63
Web design80
General162
All shared boards51101 tie
CodingLike-for-likeSix of these boards feed the Vibe Coding Index91
Coding benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
LMArena WebDev1447Not measuredOnly DeepSeek V4 Pro 0423 measured
AA Coding Index59.449.3DeepSeek V4 Pro 0423
LiveCodeBench (Vals AI)87.5%86.0%DeepSeek V4 Pro 0423
SWE-bench Verified (Vals AI)77.4%69.0%DeepSeek V4 Pro 0423
Terminal-Bench 4 (Vals AI)11.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
Vibe Code Bench (Vals AI)49.9%7.6%DeepSeek V4 Pro 0423
Code migration (Vals AI)26.2%4.9%DeepSeek V4 Pro 0423
ProgramBench (Vals AI)0.0%0.0%Tie
Terminal-Bench 2.1 (Vals AI)50.2%50.9%Nemotron 3 Ultra
SkillsBench (Vals AI)51.3%Not measuredOnly DeepSeek V4 Pro 0423 measured
SciCodeNot measured40.3%Only Nemotron 3 Ultra measured
Terminal-Bench Hard (Artificial Analysis run)Not measured36.4%Only Nemotron 3 Ultra measured
FrontierCode17.6%13.6%DeepSeek V4 Pro 0423
WeirdML48.9%43.5%DeepSeek V4 Pro 0423
ALE-Bench1006.1Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text: coding14701468DeepSeek V4 Pro 0423
LMArena Text: software and IT14691468DeepSeek V4 Pro 0423
LMArena WebDev: React1441Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena WebDev: HTML1435Not measuredOnly DeepSeek V4 Pro 0423 measured
SWE Atlas: Codebase QnA27.2%Not measuredOnly DeepSeek V4 Pro 0423 measured
SWE Atlas: Test Writing27.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 22 boards here
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
Intelligence benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
AA Intelligence Index30.422.9DeepSeek V4 Pro 0423
Epoch Capabilities Index149.1146.0DeepSeek V4 Pro 0423
Agentic workLike-for-likePublic boards, shown as published, kept out of the index40
Agentic work benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
AA Agentic Index26.320.1DeepSeek V4 Pro 0423
Design Arena: agents building web apps1000Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: agents building full-stack apps948Not measuredOnly DeepSeek V4 Pro 0423 measured
Design Arena: agents building Godot games1059Not measuredOnly DeepSeek V4 Pro 0423 measured
Finance Agent (Vals AI)44.1%37.7%DeepSeek V4 Pro 0423
Tax Agent Bench (Vals AI)25.2%Not measuredOnly DeepSeek V4 Pro 0423 measured
Legal Research Bench (Vals AI)23.1%15.4%DeepSeek V4 Pro 0423
Excel Modeling Benchmark (Vals AI)51.6%31.7%DeepSeek V4 Pro 0423
τ²-Bench TelecomNot measured83.3%Only Nemotron 3 Ultra measured
APEX-AgentsNot measured22.7%Only Nemotron 3 Ultra measured
Vending-Bench 2$3,285Not measuredOnly DeepSeek V4 Pro 0423 measured
MCP AtlasNot measured63.1%Only Nemotron 3 Ultra measured
Neither model has a result on the other 30 boards here
ReasoningLike-for-likePublic boards, shown as published, kept out of the index42
Reasoning benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
GPQA Diamond89.4%86.1%DeepSeek V4 Pro 0423
GPQA Diamond (Artificial Analysis run)Not measured86.7%Only Nemotron 3 Ultra measured
Humanity's Last Exam (Artificial Analysis run)Not measured28.4%Only Nemotron 3 Ultra measured
GPQA Diamond (Epoch AI run)90.9%85.4%DeepSeek V4 Pro 0423
Chess puzzles19.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
Mystery game puzzles17.0%20.0%Nemotron 3 Ultra
LMArena Text: hard prompts14611452DeepSeek V4 Pro 0423
LMArena Text: expert14631472Nemotron 3 Ultra
LMArena Text: hard prompts in English14661453DeepSeek V4 Pro 0423
Kagi LLM Benchmark53.5%Not measuredOnly DeepSeek V4 Pro 0423 measured
Humanity's Last Exam (Scale run)13.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 12 boards here
MathLike-for-likePublic boards, shown as published, kept out of the index22
Math benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
OTIS Mock AIME96.7%86.7%DeepSeek V4 Pro 0423
FrontierMath, tiers 1 to 345.3%Not measuredOnly DeepSeek V4 Pro 0423 measured
FrontierMath, tier 42.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
ProofBench16.0%2.0%DeepSeek V4 Pro 0423
LMArena Text: math14431457Nemotron 3 Ultra
LMArena Text: mathematical work14551461Nemotron 3 Ultra
MathArena26.2%Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 7 boards here
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index63
Knowledge benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
MMLU Pro87.3%85.8%DeepSeek V4 Pro 0423
LegalBench (Vals AI)80.3%82.1%Nemotron 3 Ultra
CaseLaw (Vals AI)59.4%Not measuredOnly DeepSeek V4 Pro 0423 measured
CorpFin (Vals AI)61.4%65.5%Nemotron 3 Ultra
TaxEval (Vals AI)72.1%73.1%Nemotron 3 Ultra
MedCode (Vals AI)40.5%38.6%DeepSeek V4 Pro 0423
SimpleQA Verified47.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text: science14701465DeepSeek V4 Pro 0423
LMArena Text: medicine and healthcare14661456DeepSeek V4 Pro 0423
LMArena Text: legal and government14641452DeepSeek V4 Pro 0423
LMArena Text: business and finance14431434DeepSeek V4 Pro 0423
Neither model has a result on the other 9 boards here
Web designLike-for-likePublic boards, shown as published, kept out of the index80
Web design benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
Design Arena: websites12341145DeepSeek V4 Pro 0423
Design Arena: UI components12191135DeepSeek V4 Pro 0423
Design Arena: game dev12281138DeepSeek V4 Pro 0423
Design Arena: data viz12031141DeepSeek V4 Pro 0423
Design Arena: 3D12531131DeepSeek V4 Pro 0423
Design Arena: all code categories12401146DeepSeek V4 Pro 0423
Design Arena: SVG11391069DeepSeek V4 Pro 0423
Design Arena: ASCII art11491100DeepSeek V4 Pro 0423
Neither model has a result on the other 1 board here
GeneralLike-for-likePublic boards, shown as published, kept out of the index162
General benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
Public Benefits Bench (Vals AI)62.9%Not measuredOnly DeepSeek V4 Pro 0423 measured
MedScribe (Vals AI)75.1%Not measuredOnly DeepSeek V4 Pro 0423 measured
IFBenchNot measured81.4%Only Nemotron 3 Ultra measured
Long context reasoning (AA-LCR)Not measured79.3%Only Nemotron 3 Ultra measured
DTBench90.7%90.1%DeepSeek V4 Pro 0423
LMCA41.2%36.9%DeepSeek V4 Pro 0423
Surface Evolver Bench40.0%Not measuredOnly DeepSeek V4 Pro 0423 measured
CL-bench Life13.5%Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text14511445DeepSeek V4 Pro 0423
LMArena Text: instruction following14451418DeepSeek V4 Pro 0423
LMArena Text: creative writing14461399DeepSeek V4 Pro 0423
LMArena Text: English14591455DeepSeek V4 Pro 0423
LMArena Text: non-English14391427DeepSeek V4 Pro 0423
LMArena Text: Chinese14811497Nemotron 3 Ultra
LMArena Text: German14581472Nemotron 3 Ultra
LMArena Text: Japanese1446Not measuredOnly DeepSeek V4 Pro 0423 measured
LMArena Text: Korean14091384DeepSeek V4 Pro 0423
LMArena Text: French14731472DeepSeek V4 Pro 0423
LMArena Text: Spanish14571455DeepSeek V4 Pro 0423
LMArena Text: Russian14531417DeepSeek V4 Pro 0423
LMArena Text: Polish14611385DeepSeek V4 Pro 0423
LMArena Text: multi-turn14681411DeepSeek V4 Pro 0423
LMArena Text: longer queries14581437DeepSeek V4 Pro 0423
LMArena Text: writing and language14471406DeepSeek V4 Pro 0423
LMArena Text: entertainment and media14311397DeepSeek V4 Pro 0423
EQ-Bench 41166Not measuredOnly DeepSeek V4 Pro 0423 measured
Neither model has a result on the other 22 boards here
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
Image generation benchmarks for DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
BenchmarkDeepSeek V4 Pro 0423Nemotron 3 UltraLeader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

DeepSeek V4 Pro 0423$0.0026Fits in one request
Nemotron 3 Ultra$0.0018Fits in one request

Nemotron 3 Ultra costs less

Repository review

50K input + 3K output tokens

DeepSeek V4 Pro 0423$0.073Fits in one request
Nemotron 3 Ultra$0.037Fits in one request

Nemotron 3 Ultra costs less

Agent loop

200K input + 10K output tokens

DeepSeek V4 Pro 0423$0.286Fits in one request
Nemotron 3 Ultra$0.144Fits in one request

Nemotron 3 Ultra costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of DeepSeek V4 Pro 0423 and Nemotron 3 Ultra
SpecificationDeepSeek V4 Pro 0423NVIDIA Nemotron 3 Ultra
MakerDeepSeekNVIDIA
WeightsOpen weightsOpen weights
ReleasedApril 24, 2026June 4, 2026
Context window1.05M262K
Input $/Mtok$1.30$0.60
Output $/Mtok$2.60$2.40
Coding index38.730.0
Independent measurements behind it53
Cost per solved task$0.57Not computable
SourceOpenRouter listingOpenRouter listing

Questions

Is DeepSeek V4 Pro 0423 better than Nemotron 3 Ultra for coding?

Pick DeepSeek V4 Pro 0423 for coding: it scores 38.7 on the Vibe Coding Index against 30.0 for Nemotron 3 Ultra. Pick Nemotron 3 Ultra for output tokens 1.1x cheaper ($2.40 vs $2.60 per million) and work measured by Terminal-Bench 2.1 (Vals AI) or Mystery game puzzles, where it leads.

Which is cheaper, DeepSeek V4 Pro 0423 or Nemotron 3 Ultra?

DeepSeek V4 Pro 0423 costs $1.30 per million input tokens and $2.60 per million output tokens; Nemotron 3 Ultra costs $0.60 and $2.40. Cost per solved task is known for DeepSeek V4 Pro 0423 only.

Which has the bigger context window, DeepSeek V4 Pro 0423 or Nemotron 3 Ultra?

DeepSeek V4 Pro 0423, with 1.05M tokens against 262K.

How many benchmarks is the DeepSeek V4 Pro 0423 vs Nemotron 3 Ultra comparison based on?

62 benchmarks have a result for both models as of October 22, 2026. DeepSeek V4 Pro 0423's coding index rests on 5 independent benchmarks and Nemotron 3 Ultra's on 3. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/deepseek-v4-pro-vs-nvidia-nemotron-3-ultra-550b-a55b" title="DeepSeek V4 Pro 0423 vs Nemotron 3 Ultra, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/deepseek-v4-pro-vs-nvidia-nemotron-3-ultra-550b-a55b">LLMBenchmark.ai, DeepSeek V4 Pro 0423 vs Nemotron 3 Ultra</a></p>