Nemotron 3 Ultra 550B A55B vs Step 3.7 Flash: Benchmarks, Pricing & Capabilities Compared
TL;DR — Nemotron 3 Ultra 550B A55B wins for reasoning + long-context · Step 3.7 Flash wins for cost.
Nemotron 3 Ultra 550B A55B NVIDIA
- Released
- 2026-06-04
- Context window
- 262K tokens
- Input price
- $0.60 / Mtok
- Output price
- $2.60 / Mtok
Step 3.7 Flash StepFun
- Released
- 2026-05-29
- Context window
- 256K tokens
- Input price
- $0.20 / Mtok
- Output price
- $1.15 / Mtok
Benchmark comparison
| Benchmark | Nemotron 3 Ultra 550B A55B | Step 3.7 Flash |
|---|---|---|
| AA Intelligence Index | 47.7 ✓ | 42.6 |
| GPQA Diamond | 86.7% ✓ | 80.9% |
| HLE | 26.6% ✓ | 19.9% |
| IF-Bench | 81.4% ✓ | 67.3% |
| LiveCodeBench Reasoning | 67.0% ✓ | 63.7% |
| SciCode | 39.9% | 40.0% ✓ |
| TAU2-bench | 83.3% | 98.5% ✓ |
| TerminalBench-Hard | 36.4% ✓ | 35.6% |
Pricing comparison
| Metric | Nemotron 3 Ultra 550B A55B | Step 3.7 Flash |
|---|---|---|
| Input ($/Mtok) | $0.60 | $0.20 |
| Output ($/Mtok) | $2.60 | $1.15 |
| Cached input ($/Mtok) | — | — |
| Cost per 1M-token roundtrip (1M in + 1M out) | $3.20 | $1.35 |
Context window & modalities
| Attribute | Nemotron 3 Ultra 550B A55B | Step 3.7 Flash |
|---|---|---|
| Context window | 262K tokens | 256K tokens |
| Input modalities | text | text, image |
| Output modalities | text | text |
| Knowledge cutoff | — | — |
Verdict by use case
Coding
Insufficient data
Basis: SWE-bench
No shared coding benchmark.
Reasoning
→ Nemotron 3 Ultra 550B A55B
Basis: GPQA Diamond
Nemotron 3 Ultra 550B A55B 86.7% vs Step 3.7 Flash 80.9% on GPQA Diamond.
Math
Insufficient data
Basis: MATH / AIME
No shared math benchmark.
Long context
→ Nemotron 3 Ultra 550B A55B
Basis: Context window
Nemotron 3 Ultra 550B A55B 262K tokens vs Step 3.7 Flash 256K tokens.
Cost
→ Step 3.7 Flash
Basis: Input $/Mtok
Nemotron 3 Ultra 550B A55B $0.6/Mtok vs Step 3.7 Flash $0.2/Mtok input.
Changelog & releases
Nemotron 3 Ultra 550B A55B
Released 2026-06-04
Step 3.7 Flash
Released 2026-05-29