Largest Nemotron 3 model for maximum open-weight reasoning and agent accuracy
Specification
most-agreed values
Context
131K
Max output
32K
Released
2026-06-04
Knowledge cutoff
2025-09
Retires
—
Open weights
yes
Input
text
Output
text
Price
US dollars per million tokens · most-agreed
Input
$0.10
Output
$0.10
Cache read
$0.10
Quality
3 benchmarks · 1 source
Composite
82nd
percentile of 344 scored models
Rank
62 / 344
±0.09 sd
Evidence
3 × 1
one source, or fewer than 3 benchmarks
Effort range
—
one configuration only
reasoning73rd2/4 bench
math89th1/7 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 86.7 ±5.1 | default | Epoch AI | 2026-08-10 |
| GPQA diamond | 85.3 ±2.5 | default | Epoch AI | 2026-08-10 |
| Mystery Game Puzzles | 20.0 ±4.0 | default | Epoch AI | 2026-08-27 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 2 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| ollama-cloud | $0.10 | $3.00 | $0.10 | — | 262K | 128K | — |
| routing-run | $0.10 | $0.10 | — | — | 131K | 32K | — |
Other listings of this model
this page is one of them
The corpus files Nemotron 3 Ultra 550B A55B under several keys. This page is the nemotron-3-ultra listing. The full record — every host, every price and every benchmark score — is on the main Nemotron 3 Ultra 550B A55B page.
More from NVIDIA
most-hosted first