Skip to main content
BenchLM
Advertise

Model comparison

DeepSeek V4 Pro (Max) vs Nemotron 3 Ultra

Data verified

Head-to-head evidence from 27 shared benchmark results across 6 categories. Overall scores shown here use the public BenchAlign v5 ranking lane.

76/100
Margin
16.0pts
← winning
60/100
3 category wins0 category wins

Verified leaderboard positions: DeepSeek V4 Pro (Max) #9; Nemotron 3 Ultra #24

Evidence parity. DeepSeek V4 Pro (Max) and Nemotron 3 Ultra share 27 comparable benchmark results. 3 of 8 categories are comparable. 15 results are unique to DeepSeek V4 Pro (Max); 12 to Nemotron 3 Ultra.

Updated July 16, 2026
Shared results
27
DeepSeek V4 Pro (Max) only
15
Nemotron 3 Ultra only
12
Comparable categories
3 / 8

Pick DeepSeek V4 Pro (Max) if you want the stronger benchmark profile. Nemotron 3 Ultra only becomes the better choice if you want the cheaper token bill.

Confidence note. This is a partial-evidence comparison with 27 shared benchmark results across 6 evidence categories; 3 of 8 categories currently have scoreable aggregates for both models. Treat the verdict as directional until coverage is more balanced.

Why this result

DeepSeek V4 Pro (Max) is clearly ahead on the provisional aggregate, 76 to 60. The gap is large enough that you do not need to squint at the spreadsheet to see the difference.

DeepSeek V4 Pro (Max)'s sharpest advantage is in agentic, where it averages 74.5 against 51.3. The single biggest benchmark swing on the page is BrowseComp, 83.4% to 44.4%.

DeepSeek V4 Pro (Max) is also the more expensive model on tokens at $0.43 input / $0.87 output per 1M tokens, versus $0.00 input / $0.00 output per 1M tokens for Nemotron 3 Ultra. That is roughly Infinityx on output cost alone.

Category breakdown

Exact category averages are shown below. Not measured means BenchLM does not have enough sourced public coverage for that model and category.

Category scores and score margins for DeepSeek V4 Pro (Max) and Nemotron 3 Ultra
CategoryDeepSeek V4 Pro (Max)ΔNemotron 3 Ultra
AgenticDeepSeek V4 Pro (Max)74.5Margin← 23.2Nemotron 3 Ultra51.3
CodingDeepSeek V4 Pro (Max)70.9Margin← 12.6Nemotron 3 Ultra58.3
KnowledgeDeepSeek V4 Pro (Max)60.4Margin← 6.2Nemotron 3 Ultra54.2
ReasoningDeepSeek V4 Pro (Max)Not measuredMarginNo overlapNemotron 3 Ultra61.9
MathDeepSeek V4 Pro (Max)95.2MarginNo overlapNemotron 3 UltraNot measured
MultilingualDeepSeek V4 Pro (Max)Not measuredMarginNo overlapNemotron 3 Ultra83.0
Inst. FollowingDeepSeek V4 Pro (Max)Not measuredMarginNo overlapNemotron 3 Ultra81.7

Decisive benchmark drivers

The largest measured benchmark gaps in this matchup, with exact reported values.

More
A · DeepSeek V4 Pro (Max)B · Nemotron 3 Ultra

  1. BrowseComp

    Agentic
    Source ↗
    A 83.4%B 44.4%
    Winner: DeepSeek V4 Pro (Max)Δ 39
    BrowseComp: DeepSeek V4 Pro (Max) scored 83.4%; Nemotron 3 Ultra scored 44.4%. DeepSeek V4 Pro (Max) wins this benchmark.

  2. Terminal-Bench 2.0

    Agentic
    Source ↗
    A 67.9%B 56.4%
    Winner: DeepSeek V4 Pro (Max)Δ 11.5
    Terminal-Bench 2.0: DeepSeek V4 Pro (Max) scored 67.9%; Nemotron 3 Ultra scored 56.4%. DeepSeek V4 Pro (Max) wins this benchmark.

  3. HLE

    Knowledge
    Source ↗
    A 37.7%B 26.7%
    Winner: DeepSeek V4 Pro (Max)Δ 11
    HLE: DeepSeek V4 Pro (Max) scored 37.7%; Nemotron 3 Ultra scored 26.7%. DeepSeek V4 Pro (Max) wins this benchmark.

  4. SWE-bench Verified

    Coding
    Source ↗
    A 80.6%B 71.9%
    Winner: DeepSeek V4 Pro (Max)Δ 8.7
    SWE-bench Verified: DeepSeek V4 Pro (Max) scored 80.6%; Nemotron 3 Ultra scored 71.9%. DeepSeek V4 Pro (Max) wins this benchmark.

  5. GPQA

    Knowledge
    Source ↗
    A 90.1%B 87%
    Winner: DeepSeek V4 Pro (Max)Δ 3.1
    GPQA: DeepSeek V4 Pro (Max) scored 90.1%; Nemotron 3 Ultra scored 87%. DeepSeek V4 Pro (Max) wins this benchmark.

Operational comparison

Runtime and commercial metrics are compared only when both models have a complete sourced value.

MetricDeepSeek V4 Pro (Max)Nemotron 3 UltraComparison
Input / output priceUSD per 1M tokensDeepSeek V4 Pro (Max)$0.435 input / $0.87 outputNemotron 3 Ultra$0 input / $0 outputNemotron 3 Ultra has the lower combined listed price.
Generation speedtokens per secondDeepSeek V4 Pro (Max)Not availableNemotron 3 UltraNot availableA complete speed comparison is not available.
First-answer latencyseconds to first tokenDeepSeek V4 Pro (Max)Not availableNemotron 3 UltraNot availableA complete latency comparison is not available.
Context windowmaximum listed tokensDeepSeek V4 Pro (Max)1MNemotron 3 Ultra1MListed context windows are equal.

Benchmark Deep Dive

AgenticDeepSeek V4 Pro (Max) wins
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
Terminal-Bench 2.0Source 67.9%56.4%DeepSeek V4 Pro (Max) leads
BrowseCompSource 83.4%44.4%DeepSeek V4 Pro (Max) leads
HLE w/ toolsSource 48.2%37.4%DeepSeek V4 Pro (Max) leads
MCP AtlasSource 73.6%—Not comparable
GDPval-AASource 13071164DeepSeek V4 Pro (Max) leads
ToolathlonSource 51.8%—Not comparable
AA Agentic IndexSource 36.4%27.4%DeepSeek V4 Pro (Max) leads
APEX-Agents-AASource 24.3%—Not comparable
τ²-bench resultsSource 96.2%83.3%DeepSeek V4 Pro (Max) leads
GDPval-AASource 40.4%33.2%DeepSeek V4 Pro (Max) leads
PinchBenchSource —90.0%Not comparable
τ³-bench resultsSource —70.9%Not comparable
AA BriefcaseSource —870Not comparable
AA EnterpriseOps-GymSource —28.9%Not comparable
AA Harvey LABSource —3.3%Not comparable
CodingDeepSeek V4 Pro (Max) wins
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
LiveCodeBench Pass@1-COTSource 93.5%—Not comparable
CodeforcesSource 3206.0—Not comparable
SWE-bench VerifiedSource 80.6%71.9%DeepSeek V4 Pro (Max) leads
SWE-bench ProSource 55.4%—Not comparable
SWE MultilingualSource 76.2%67.7%DeepSeek V4 Pro (Max) leads
Terminal-Bench 2.0Source 67.9%56.4%DeepSeek V4 Pro (Max) leads
Vibe Code BenchSource 49.93%—Not comparable
AA Coding IndexSource 59.4%49.3%DeepSeek V4 Pro (Max) leads
Terminal-Bench HardSource 46.2%36.4%DeepSeek V4 Pro (Max) leads
AA-SciCodeSource 50.0%39.9%DeepSeek V4 Pro (Max) leads
LiveCodeBench v6Source —89.0%Not comparable
SciCodeSource —44.6%Not comparable
Reasoning
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
MRCR 1MSource 83.5%—Not comparable
CorpusQA 1MSource 62.0%—Not comparable
AA-LCRSource 66.3%67.0%Nemotron 3 Ultra leads
CritPtSource 12.9%3.1%DeepSeek V4 Pro (Max) leads
LongBench v2Source —61.9%Not comparable
KnowledgeDeepSeek V4 Pro (Max) wins
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
MMLU-ProSource 87.5%86.8%DeepSeek V4 Pro (Max) leads
SimpleQASource 57.9%—Not comparable
Chinese-SimpleQASource 84.4%—Not comparable
GPQASource 90.1%87%DeepSeek V4 Pro (Max) leads
GPQA-DSource 90.1%87.0%DeepSeek V4 Pro (Max) leads
HLESource 37.7%26.7%DeepSeek V4 Pro (Max) leads
Artificial Analysis Intelligence IndexSource 44.3%37.8%DeepSeek V4 Pro (Max) leads
AA-GPQA DiamondSource 88.8%86.7%DeepSeek V4 Pro (Max) leads
AA-HLESource 35.9%26.6%DeepSeek V4 Pro (Max) leads
AA-Omniscience IndexSource -10.0%-0.8%Nemotron 3 Ultra leads
AA-Omniscience AccuracySource 43.3%21.6%DeepSeek V4 Pro (Max) leads
AA-Omniscience Hallucination RateSource 94.0%28.5%Nemotron 3 Ultra leads
HLE w/o toolsSource —26.7%Not comparable
AA Openness IndexSource —83.3%Not comparable
Math
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
HMMT Feb 2026Source 95.2%—Not comparable
IMOAnswerBenchSource 89.8%—Not comparable
ApexSource 38.3%—Not comparable
Apex ShortlistSource 90.2%—Not comparable
Multilingual
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
MMLU-ProXSource —83%Not comparable
Multimodal
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
Design Arena WebsiteSource 12691132DeepSeek V4 Pro (Max) leads
Inst. Following
BenchmarkDeepSeek V4 Pro (Max)Nemotron 3 UltraResult
AA-IFBenchSource 76.5%81.4%Nemotron 3 Ultra leads
IFBenchSource —81.7%Not comparable
Frequently Asked Questions (4)

Which is better, DeepSeek V4 Pro (Max) or Nemotron 3 Ultra?

DeepSeek V4 Pro (Max) is ahead on BenchLM's provisional leaderboard, 76 to 60. The biggest single separator in this matchup is BrowseComp, where the scores are 83.4% and 44.4%.

Which is better for knowledge tasks, DeepSeek V4 Pro (Max) or Nemotron 3 Ultra?

DeepSeek V4 Pro (Max) has the edge for knowledge tasks in this comparison, averaging 60.4 versus 54.2. Inside this category, AA-Omniscience Hallucination Rate is the benchmark that creates the most daylight between them.

Which is better for coding, DeepSeek V4 Pro (Max) or Nemotron 3 Ultra?

DeepSeek V4 Pro (Max) has the edge for coding in this comparison, averaging 70.9 versus 58.3. Inside this category, Terminal-Bench 2.0 is the benchmark that creates the most daylight between them.

Which is better for agentic tasks, DeepSeek V4 Pro (Max) or Nemotron 3 Ultra?

DeepSeek V4 Pro (Max) has the edge for agentic tasks in this comparison, averaging 74.5 versus 51.3. Inside this category, GDPval-AA is the benchmark that creates the most daylight between them.

Related Comparisons

Explore More

DeepSeek Compare Pricing Methodology Find Your Best LLM Overall Rankings
Last updated: July 16, 2026

The AI models change fast. We track them for you.

A weekly brief for engineers and researchers covering new models, ranking shifts, and pricing changes.

Free. No spam. Unsubscribe anytime.