Your sovereign, private Nemotron Lightning 3.5 30B
Nemotron-Lightning-3.5-30B-A3B is a 30B-parameter MoE (3B active) from NVIDIA's Nemotron-H family, built on a hybrid Mamba-Transformer architecture for efficient long-context inference. It generates a reasoning trace before its final response (configurable via chat template) and ships with Multi-Token Prediction speculative decoding for low-latency serving. A reasoning model in the small tier, served on ToothFairyAI's AU, EU and US endpoints — private by default, billed in US dollars per million tokens at the published rates below. No training on your data, no vendor lock-in.
How Nemotron Lightning 3.5 30B compares
Independent intelligence index for Nemotron Lightning 3.5 30B against every peer in the Small tier (and the tier above where needed) — same source, same tests.
Throughput — tokens per second, median
Latency — end-to-end seconds for a 500-token answer
Capability indexes
Domain-weighted agentic performance for Nemotron Lightning 3.5 30B.
Capability index vs small peers
The evaluations behind the index
Independent benchmark results for Nemotron Lightning 3.5 30B — Elo scores as published; pass rates shown as percentages.
Per-evaluation economics
| Evaluation | Score | Time / task | Output tokens / task |
|---|---|---|---|
| AA-Briefcase | 506.17 | 4.0 min | 80038 |
| GDPval-AA | 638.57 | 2.5 min | 50303 |
| AutomationBench-AA | 0.8% | 1.0 min | 19152 |
| Terminal-Bench 4.0 | 0.5% | 4.2 min | 84950 |
| SciCode | 32.1% | 0.7 min | 14828 |
| Humanity's Last Exam | 10.6% | 1.3 min | 26623 |
| GDP.pdf | 3.0% | 0.7 min | 13731 |
| CritPt | 0.0% | 2.0 min | 39844 |
| AA-Omniscience | -1771.7% | 0.1 min | 2964 |
| AA-LCR (long context) | 60.3% | 0.4 min | 7255 |
Run Nemotron Lightning 3.5 30B privately
One key, one endpoint, three jurisdictions.
- Size tierSmall
- Serving routeStandard
- Input rate$0.17/1M
- Output rate$0.51/1M
- Cached input rate$0.09/1M
- Context window262K tokens
- Max output16K tokens
- ReasoningYes — emits thinking traces
- Vision & video inputText only
- Tool callingSupported
- Data residencyAU, EU and US endpoints — processing stays inside the region you call