Back to model shop
Nvidia
Rendered from D1 model snapshot · D1 read 90msCache check: checking

nvidia/llama-3.1-nemotron-ultra-253b

Open weights
Llama 3.1 Nemotron Ultra 253B

Flagship Nemotron model for high-throughput reasoning and complex agents

This model's cache

This model page returns the exact headers below. The page and /api/models/nvidia--llama-3.1-nemotron-ultra-253b share the model:nvidia--llama-3.1-nemotron-ultra-253b tag, so invalidation stays scoped to this model.

Cache key/models/nvidia--llama-3.1-nemotron-ultra-253b
Cache-Controlpublic, max-age=300, stale-while-revalidate=3600
Cache-Tagcatalogmodel:nvidia--llama-3.1-nemotron-ultra-253b

Workers Cache derives the key from this path/query, plus the Worker entrypoint and current Worker version.

export const Route = createFileRoute('/models/$modelSlug')({
  headers: () => ({
    'Cache-Control': 'public, max-age=300, stale-while-revalidate=3600',
    'Cache-Tag': 'catalog,model:nvidia--llama-3.1-nemotron-ultra-253b',
  }),
})

Invalidate this cache tag

await ctx.cache.purge({ tags: ['model:nvidia--llama-3.1-nemotron-ultra-253b'] })

Use this when one model changes and you do not want to evict the whole catalog.

NemotronReasoningTool callingTemperature control
Released
Apr 7, 2025
Last updated
Apr 7, 2025
Context window
128K tokens
Max output
8.2K tokens
Open weights
Yes
Family
Nemotron

Starts at

Quote

per 1M input tokens

Output

Quote

Sellers

0

No cart or checkout in this MVP — pick a provider below and open their docs when you're ready to use it.

Capabilities & modalities
Reasoning
Tool calling
Structured output
Attachments
InputText
OutputText
Provider shelf
Where to get Llama 3.1 Nemotron Ultra 253B

Sorted by lowest input price

Models.dev has metadata for this model, but this MVP snapshot did not find a matching provider offer.

More from Nvidia