Model
Nemotron 3 Ultra 550B A55B
Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy
6 providersreleased open weights
$0.5
best input /M · Kilo Gateway
$2.20
best output /M
1M
context window
128K
max output
Compare providers (sorted by listed input price)
| Provider | Input /M | Output /M | Cache read /M | Cache write /M | Context | Capabilities | Status | Updated |
|---|---|---|---|---|---|---|---|---|
| Nvidia nvidia/nemotron-3-ultra-550b-a55b | $0.5 | $2.50 | $0.15 | — | 1M | reasoningtoolsstructuredno vision | Jun 4, 2026 | |
| NanoGPT nvidia/nemotron-3-ultra-550b-a55b | $0.5 | $2.50 | $0.25 | — | 1M | reasoningtoolsno structuredno vision | Jun 4, 2026 | |
| Kilo Gateway nvidia/nemotron-3-ultra-550b-a55b | $0.5 | $2.20 | $0.1 | — | 512K | reasoningtoolsstructuredno vision | Jun 4, 2026 | |
| OpenRouter nvidia/nemotron-3-ultra-550b-a55b | $0.6 | $3.60 | $0.2 | — | 512K | reasoningtoolsstructuredno vision | Jun 4, 2026 | |
| Together AI nvidia/nemotron-3-ultra-550b-a55b | $0.6 | $3.60 | $0.2 | — | 512K | reasoningtoolsstructuredno vision | Jun 4, 2026 | |
| Vercel AI Gateway nvidia/nemotron-3-ultra-550b-a55b | $0.6 | $2.40 | $0.12 | — | 1M | reasoningtoolsno structuredno vision | Jun 4, 2026 |
Price badge
https://models.sutraworks.ai/badge/nvidia/nemotron-3-ultra-550b-a55b.svgLive SVG, regenerated on every hourly sync — paste it into a README to always show the current cheapest listed price. Updates when prices change; no build step needed on your side.
Benchmarks
- SWE-Bench Verified ↗(resolved)report ↗70.7
- SWE-Bench Multilingual ↗(resolve rate)report ↗67.7
- Terminal-Bench ↗(success rate)report ↗56.4
- GPQA ↗(accuracy)report ↗87
- Humanity's Last Exam ↗(accuracy)report ↗26.7
- Humanity's Last Exam ↗(accuracy)report ↗37.4
- LiveCodeBench ↗(pass@1)report ↗89
- MMLU-Pro ↗(accuracy)report ↗86.8
- BrowseComp ↗(accuracy)report ↗44.4
- IFBench(accuracy)report ↗81.7
- GDPval(wins or ties)report ↗46.7
Prices are per million tokens (USD). “—” means the provider does not publicly list a price for this model. Data from models.dev; verify with the provider before purchasing.