Nemotron 3 Ultra (550B A55B) vs Nemotron 3.5 Lightning (30B A3B): Specs & Benchmark Comparison

Nemotron 3 Ultra (550B A55B) and Nemotron 3.5 Lightning (30B A3B) both come from NVIDIA, Nemotron 3 Ultra (550B A55B) was released in June 2026, and Nemotron 3.5 Lightning (30B A3B) followed 2 months later in August 2026. Nemotron 3 Ultra (550B A55B) is the larger model at roughly 550 billion parameters, against 30 billion for Nemotron 3.5 Lightning (30B A3B).

The two models share 12 published benchmarks. Nemotron 3 Ultra (550B A55B) leads on 12 of them. The widest gaps are on OmniScience, where Nemotron 3 Ultra (550B A55B) scores 78.7% against 17.5%; Terminal-Bench 2.1, where Nemotron 3 Ultra (550B A55B) scores 56.4% against 24.6%. Averaged across everything we track, Nemotron 3 Ultra (550B A55B) sits at 64.8% and Nemotron 3.5 Lightning (30B A3B) at 45.3%.

Nemotron 3.5 Lightning (30B A3B) is available through an API at $0.05 per million input tokens with a 262K-token context window. We do not track a hosted API for Nemotron 3 Ultra (550B A55B), so pricing and context cannot be compared directly.

CharacteristicNemotron 3 Ultra (550B A55B)Nemotron 3.5 Lightning (30B A3B)
CompanyNVIDIANVIDIA
Release DateJune 4, 2026August 11, 2026
Parameters550B30B
MultimodalNoNo
Context (input)262K
Context (output)262K
Input Price / 1M$0.05
Output Price / 1M$0.20
Average Score64.8%45.3%
Benchmarks
OmniScience78.7%17.5%
Terminal-Bench 2.156.4%24.6%
SWE-bench Multilingual67.7%39.3%
Humanity's Last Exam37.4%11.7%
SWE-Bench Verified70.7%51.6%
AA-LCR65.4%52.0%
GPQA87.0%75.0%
SciCode44.6%32.6%
IFBench81.7%72.0%
BrowseComp44.4%37.0%
PinchBench90.0%85.0%
MMLU-Pro86.8%82.0%

Visual Benchmark Comparison

Nemotron 3 Ultra (550B A55B)
Nemotron 3.5 Lightning (30B A3B)
OmniScience0.8 vs 0.2
0.8
0.2
Terminal-Bench 2.10.6 vs 0.2
0.6
0.2
SWE-bench Multilingual0.7 vs 0.4
0.7
0.4
Humanity's Last Exam0.4 vs 0.1
0.4
0.1
SWE-Bench Verified0.7 vs 0.5
0.7
0.5
AA-LCR0.7 vs 0.5
0.7
0.5
GPQA0.9 vs 0.8
0.9
0.8
SciCode0.4 vs 0.3
0.4
0.3
IFBench0.8 vs 0.7
0.8
0.7
BrowseComp0.4 vs 0.4
0.4
0.4
PinchBench0.9 vs 0.8
0.9
0.8
MMLU-Pro0.9 vs 0.8
0.9
0.8

Verdict

Nemotron 3.5 Lightning (30B A3B) leads in 1 out of 3 comparison categories.

Overall Performance

Both models show comparable average scores: Nemotron 3 Ultra (550B A55B) — 0.6, Nemotron 3.5 Lightning (30B A3B) — 0.5.

Programming

On SWE-Bench, both models are nearly equal: Nemotron 3 Ultra (550B A55B) — 0.7, Nemotron 3.5 Lightning (30B A3B) — 0.5.

Recency

Nemotron 3.5 Lightning (30B A3B) is newer: released 8/11/2026 vs 6/4/2026.

More About These Models

Related Comparisons

Frequently Asked Questions

Which is better for coding — Nemotron 3 Ultra (550B A55B) or Nemotron 3.5 Lightning (30B A3B)?
On the SWE-Bench benchmark, Nemotron 3 Ultra (550B A55B) shows a better result: 70.7% vs 51.6%.
Which model is cheaper — Nemotron 3 Ultra (550B A55B) or Nemotron 3.5 Lightning (30B A3B)?
API pricing data is available on the individual model pages.
Which has a larger context window — Nemotron 3 Ultra (550B A55B) or Nemotron 3.5 Lightning (30B A3B)?
Context window data is available on the individual model pages.

The Nemotron 3 Ultra (550B A55B) and Nemotron 3.5 Lightning (30B A3B) comparison is updated for 2026. Data includes benchmark results, API pricing, context window size and other specifications. For more detailed information, visit the Nemotron 3 Ultra (550B A55B) or Nemotron 3.5 Lightning (30B A3B) page. See also the complete list of AI model comparisons.