Skip to contentT/TTokens or Towers

Compare

LeaderboardEvery tracked model, ranked and priced side by sideBenchmarksWhat each test measures and who leads itPricesAPI, media, and hardware rates with a cost calculator

Plan

Stack plannerSix questions to a priced local and hosted stackShortlistOur picks for subscriptions, APIs, and hardwareEnterpriseSeat plans against self-hosting for a whole team

Build

HardwareGPUs and workstations, and which models they fitHostsWhere to rent open-weight inference, and for how muchHarnessesCoding agents and the models they drive

Lab

FindingsResults from our own evaluationsScanAudit a repo’s AI usage in the browserMethodWhere the numbers come from and how we check them
Data checked September 25, 2026
Data checked September 25, 2026
Plan a stack
T/TTokens or Towers

An independent guide to AI models, what they cost, and the hardware to run them. No account, no API key, no sponsors.

Compare

  • Leaderboard
  • Benchmarks
  • Prices

Plan

  • Stack planner
  • Shortlist
  • Enterprise

Build

  • Hardware
  • Hosts
  • Harnesses

Lab

  • Findings
  • Scan
  • Method
Data checked September 25, 2026 · Every number links to its sourceRSS · Source on GitHub
Leaderboard/NVIDIA
NVIDIA

NVIDIA · Nemotron 3

Nemotron 3 Super 120B A12B

Open weightsGenerally availableNVIDIA Nemotron Open Model License

Mid-size open hybrid Mamba-MoE (120B total / 12B active) for agentic reasoning with a 1M context window.

AA Intelligence
12.8#86 of 142
Input / output per 1M
$0.30/ $0.90
Output speed
169tok/s
Context window
1M tokens

Composite indexes

  • Artificial Analysis Intelligence Index#86 of 142
    13independentIntelligence Index v4.3; reasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: Claude Opus 5.5 at 58

Coding

  • SciCode#81 of 94
    36.2%independentreasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: Claude Opus 5.5 at 66.9%

  • Terminal-Bench 4.0#82 of 88
    0%independentreasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: Claude Mythos 5.1 at 60.9%

  • Terminal-Bench 2.x#73 of 103
    39.6%vendorTerminal-Bench 2.1; NVIDIA harnessNVIDIA Nemotron 3.5 Lightning model card (NVIDIA-measured comparison table) ↗
    38.6%independentTerminal-Bench 2.1; reasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: Claude Fable 5.1 at 91.4%

  • SWE-bench Verified#11 of 16
    63.1%vendorNVIDIA harnessNVIDIA Nemotron 3.5 Lightning model card (NVIDIA-measured comparison table) ↗

    Leader: Claude Sonnet 5 at 85.2%

Agents and tool use

  • τ²-Bench#38 of 88
    67.8%independenttelecom; reasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: GLM-5.2 at 99.1%

  • GDPval-AA#78 of 103
    477independentreasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: Claude Opus 5.5 at 1846

  • BrowseComp#14 of 14
    22.8%vendorNVIDIA harnessNVIDIA Nemotron 3.5 Lightning model card (NVIDIA-measured comparison table) ↗

    Leader: GPT-6 Astra at 91.5%

Reasoning and knowledge

  • GPQA Diamond#72 of 142
    80%independentreasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗
    78.6%vendorno tools; NVIDIA harnessNVIDIA Nemotron 3.5 Lightning model card (NVIDIA-measured comparison table) ↗

    Leader: GPT-6 Astra at 96.1%

  • Humanity’s Last Exam#72 of 147
    20.8%independentno tools; reasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗
    20.3%vendortext-only, no tools; NVIDIA harnessNVIDIA Nemotron 3.5 Lightning model card (NVIDIA-measured comparison table) ↗

    Leader: Claude Opus 5.5 at 61.4%

  • CritPt#58 of 139
    3.1%independentreasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: GPT-5.6 Sol at 32.3%

  • MMLU-Pro#4 of 14
    83.9%vendorNVIDIA harnessNVIDIA Nemotron 3.5 Lightning model card (NVIDIA-measured comparison table) ↗

    Leader: Nemotron 3 Ultra 550B A55B at 86.8%

Vision and long context

  • AA Long Context Reasoning#78 of 139
    65.7%independentreasoningArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

    Leader: Kimi K3 at 88.7%

Other published results

ResultScoreReported bySource
Terminal-Bench Hard (AA) · reasoning28.8%independentArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗
IFBench · reasoning71.5%independentArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗
tau2-Bench Banking (AA) · reasoning10.3%independentArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗
AA-Omniscience Index · reasoning-41.5 indexindependentArtificial Analysis: Nemotron 3 Super 120B A12B (Reasoning) ↗

Specs

Released
March 11, 2026
Parameters
120B total, 12B active
Size class
100 to 500B
Input
text
Output
text
Context
1M tokens
Max output
Not published
Cached input
$0.15 / 1M
Blended 70/30
$0.48 / 1M
First token
1.45s
  • Price: Artificial Analysis model page ↗ No first-party paid NVIDIA API (build.nvidia.com is a free trial endpoint). Value is the Artificial Analysis median across third-party API providers.
  • Speed: Artificial Analysis leaderboard (Nemotron 3 Super 120B A12B (Reasoning), median across providers or first-party as shown) ↗

Will it fit your budget?

The planner prices this model against local hardware and other APIs for your workload.

Plan a stack →

Compare with

  • NVIDIANemotron 3.5 Lightning 30B A3B12.9
  • CohereCommand A+13.1
  • InceptionMercury 2.512.3
  • AmazonAmazon Nova 2 Lite13.4