NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 vs Step-5-Preview

Specs, VRAM requirements and download trends — updated 24 October 2026.

nvidia · A · Open weights

Params

560.5 B

Context

262,144

30d

1.1 M

stepfun-ai · B · Open weights

Params

604.3 B

Context

—

30d

2 K

Specification comparison

Differences are highlighted; identical values are muted.

Specification	NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16	Step-5-Preview
Deployment	Open weights · also via API	Open weights · also via API
Parameters	560.5 B	604.3 B
Architecture	NemotronHForCausalLM	Step5ForConditionalGeneration
Context length	262,144	1,000,000
Licence	other	other
Commercial use	Unknown	Unknown
Languages	12	—
Downloads 30d	1,067,855	2,467
Downloads all time	2.1 M	2 K
Quantizations on the Hub	SAFETENSORS	SAFETENSORS
Gated	No	No
First seen on the Hub	2026-06-03	2026-09-20
API input price / 1M tokens	$0.60	$1.00
API output price / 1M tokens	$2.40	$2.70
Est. cost per 1M input + 1M output tokens	$3.00	$3.70
Cached input / 1M tokens	$0.12	$0.10
AA Intelligence Index	22.9	43.7
AA Coding Index	49.3	—
AA Agentic Index	20.1	—
Max output tokens	182,520	64,000
API providers	3	4
Tool calling	Yes	Yes
Reasoning mode	Yes	Yes