NVIDIA AI & NIMopen-weight
Released: 2025-04-10

Nemotron 3 Ultra 550B

NVIDIA 550B MoE workhorse ranking among the top models on OpenRouter, delivering high-speed inference across 8x H100 clusters.

Intelligence Index
91.2/ 100
Calibrated multi-domain composite
Throughput Speed
85tok/sec
Output streaming throughput
Time To First Token
340ms
Initial chunk server latency
LiveBench Contam-Free
78.5%
September 2026 suite score

Sourced Benchmark Evaluations (61)

Verified performance across authoritative benchmarks with provenance tracking.
BenchmarkStatusScoreTrustDateSource TypeProvenance
ARC-AGI-3Active59.2%
93
2025-05-15independentSource ↗
ARC-AGI-2Nearing Saturation71%
75
2025-05-08independentSource ↗
FrontierMathActive25.3%
98
2025-05-05independentSource ↗
PutnamBenchActive47.1%
90
2025-05-04independentSource ↗
SWE-LancerDeprecated56.8%
20
2025-05-02independentSource ↗
Omni-MATHDeprecated67%
10
2025-05-02independentSource ↗
SoohakActive68.8%
98
2025-04-30independentSource ↗
FrontierCodeActive48.4%
98
2025-04-30independentSource ↗
Humanity's Last ExamActive40%
93
2025-04-28independentSource ↗
StructFlowBenchNearing Saturation71.3%
75
2025-04-28independentSource ↗
AIMENearing Saturation86%
75
2025-04-28independentSource ↗
OlympiadBenchActive74.2%
80
2025-04-27independentSource ↗
MathArenaActive72.4%
99
2025-04-26independentSource ↗
FireBenchActive70.5%
95
2025-04-26independentSource ↗
Terminal-BenchActive58.5%
80
2025-04-26independentSource ↗
OpenAI-MRCRActive76.6%
80
2025-04-25independentSource ↗
SimpleQAActive47.4%
90
2025-04-25independentSource ↗
CodeContestsDeprecated54.3%
10
2025-04-25independentSource ↗
ComplexBenchNearing Saturation68.8%
80
2025-04-24independentSource ↗
SWE-bench ProActive51.8%
90
2025-04-24independentSource ↗
LongBench-ProActive74.8%
80
2025-04-24independentSource ↗
ZeroSCROLLSNearing Saturation73%
78
2025-04-23independentSource ↗
MultiPL-EDeprecated76.8%
10
2025-04-23independentSource ↗
LiveCodeBenchSaturated68.5%
39
2025-04-22independentSource ↗
AgentIFNearing Saturation57.1%
80
2025-04-22independentSource ↗
MATHSaturated88.7%
25
2025-04-22independentSource ↗
InfiniteBenchNearing Saturation77.5%
65
2025-04-22independentSource ↗
MercuryDeprecated73.5%
10
2025-04-21independentSource ↗
SCROLLSNearing Saturation73%
65
2025-04-21independentSource ↗
MMLU-ProNearing Saturation83.9%
75
2025-04-21independentSource ↗
RULERNearing Saturation83.9%
84
2025-04-20independentSource ↗
TabMWPSaturated83.3%
25
2025-04-20independentSource ↗
CyberSecEvalDeprecated75.2%
10
2025-04-20independentSource ↗
SWE-bench VerifiedSaturated65.1%
25
2025-04-19independentSource ↗
MMLU-ReduxSaturated89.4%
25
2025-04-19independentSource ↗
QuALITYNearing Saturation80.3%
78
2025-04-19independentSource ↗
LongBenchNearing Saturation78.4%
65
2025-04-18independentSource ↗
SWE-benchDeprecated56.8%
10
2025-04-18independentSource ↗
MGSMNearing Saturation86%
65
2025-04-18independentSource ↗
AGIEvalSaturated78.3%
25
2025-04-17independentSource ↗
QASPERNearing Saturation75.7%
65
2025-04-17independentSource ↗
TruthfulQASaturated80.3%
25
2025-04-17independentSource ↗
GSM-HardDeprecated86.9%
10
2025-04-16independentSource ↗
SuperGLUESaturated92.8%
25
2025-04-16independentSource ↗
CodeXGLUEDeprecated78.5%
23
2025-04-15independentSource ↗
MT-BenchNearing Saturation8.6%
75
2025-04-15independentSource ↗
BIG-Bench HardSaturated89.2%
25
2025-04-15independentSource ↗
WinoGrandeSaturated94.6%
28
2025-04-14independentSource ↗
MBPP+Nearing Saturation85.2%
65
2025-04-14independentSource ↗
IFEvalDeprecated82.2%
25
2025-04-14independentSource ↗
GPQANearing Saturation80.3%
78
2025-04-14independentSource ↗
Needle-in-a-HaystackSaturated98.2%
44
2025-04-13independentSource ↗
HumanEval+Saturated90.2%
25
2025-04-13independentSource ↗
CommonsenseQASaturated95.6%
28
2025-04-13independentSource ↗
MBPPSaturated88.5%
25
2025-04-12independentSource ↗
GSM8KSaturated97.7%
25
2025-04-12independentSource ↗
GLUESaturated98.3%
28
2025-04-12independentSource ↗
MMLUSaturated93%
25
2025-04-11independentSource ↗
HumanEvalSaturated95.2%
25
2025-04-11independentSource ↗
ARC-AGISaturated85.5%
25
2025-04-11independentSource ↗
LiveBench (Continuous)Active78.5%
94
2025-02-01independentSource ↗
Local Hardware Execution (Ollama)Hardware Compatible
# 1. Pull and execute model locally with Ollama
ollama run nvidia-nemotron-3-ultra-550b

# 2. Or invoke via local OpenAI-compatible endpoint
curl http://localhost:11434/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nvidia-nemotron-3-ultra-550b",
    "messages": [{"role": "user", "content": "Analyze reasoning chains on AIME 2026"}]
  }'

Evaluation & Sourcing Notes

Scores listed for Nemotron 3 Ultra 550B represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.

All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.