fix: add Bedrock models nvidia.nemotron-nano-3-30b

[github-actions]

fix: add Bedrock models nvidia.nemotron-nano-3-30b

Closes #698

Source issue: #698

Summary

Field	Value
Provider	bedrock
Primary model	nvidia.nemotron-nano-3-30b
Changed models	nvidia.nemotron-nano-3-30b
Added models	nvidia.nemotron-nano-3-30b
Updated models	None
Verification sources	1 2 3 4

Verified metadata

Model	Display name	Parent	Providers	Format	Flavor	Token limits	Pricing	Lifecycle
nvidia.nemotron-nano-3-30b	NVIDIA Nemotron 3 Nano 30B		bedrock	converse	chat	input=256000, output=8192	n/a	multimodal=false

Verification notes

Verification checklist

• Model ID confirmed in >=2 independent official sources (model card + models-at-a-glance)
• Model ID format matches Bedrock convention (provider.model)
• Not already present in model_list.json (grep for nemotron-nano returns no matches for Bedrock)
• Standard/on-demand tier confirmed
• Pricing not publicly listed on Bedrock pricing page for NVIDIA models — omitted

Verification

Official sources and fields verified

1. https://docs.aws.amazon.com/bedrock/latest/userguide/model-card-nvidia-nemotron-nano-3-30b.html (AWS Bedrock model card)

   • format: converse — Converse API is supported per the model card's APIs table
   • flavor: chat — text input / text output, supports Chat Completions and Converse APIs; matches the existing nvidia.nemotron-super-3-120b entry which also uses chat
   • max_input_tokens: 256000 — model card states "Context window: 256K tokens"
   • max_output_tokens: 8192 — model card states "Max output tokens: 8K"
   • multimodal: false — model card explicitly marks Image, Audio, Video, Speech inputs as unsupported
   • available_providers: ["bedrock"] — Bedrock on-demand (Standard tier) confirmed
   • displayName: "NVIDIA Nemotron 3 Nano 30B" — follows the naming pattern of the existing sibling entry nvidia.nemotron-super-3-120b which uses "NVIDIA Nemotron 3 Super 120B"
   • Lifecycle: Active, launched Dec 15 2025, EOL date N/A — no deprecation fields needed

2. https://docs.aws.amazon.com/bedrock/latest/userguide/model-cards.html (models-at-a-glance)

   • Confirms model ID nvidia.nemotron-nano-3-30b is listed under the NVIDIA provider

Pricing

• Pricing is not publicly listed on the Bedrock pricing page for NVIDIA models. Omitted from the spec. This is consistent with the existing nvidia.nemotron-super-3-120b entry which also has no pricing fields.

sync_models (LiteLLM) cross-check

• No entry found in LiteLLM's model_prices_and_context_window_backup.json for nvidia.nemotron-nano-3-30b or any Bedrock Nemotron Nano 30B variant (also no entry for nvidia.nemotron-super-3-120b). Since sync_models has no data for this model, there are no deviations to document.

Fields not published / not applicable

• input_cost_per_mil_tokens / output_cost_per_mil_tokens: not published on Bedrock pricing page for NVIDIA models — omitted
• input_cache_read_cost_per_mil_tokens / input_cache_write_cost_per_mil_tokens: not applicable — omitted
• parent: not applicable — this is a base model, not a dated snapshot or variant of another catalog entry
• reasoning / reasoning_budget: not applicable — no reasoning/thinking features documented
• supported_regions: not applicable — this is a Bedrock model, not a Vertex model
• locations: not required for standard Bedrock models

sync_models vs proposed update

sync_models cross-check found differences. Official provider verification was used for the applied values, and sync_models discrepancies are listed below for review.

Model	Field	Proposed update	sync_models	sync_models source models
nvidia.nemotron-nano-3-30b	max_input_tokens	256000	262144	nvidia.nemotron-nano-3-30b
nvidia.nemotron-nano-3-30b	input_cost_per_mil_tokens	n/a	0.06	nvidia.nemotron-nano-3-30b
nvidia.nemotron-nano-3-30b	output_cost_per_mil_tokens	n/a	0.24	nvidia.nemotron-nano-3-30b

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
ai-proxy	Ready	Preview, Comment	May 31, 2026 8:32pm

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: ef9e82e01a

ℹ️ About Codex in GitHub

Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "Codex (@codex) review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "Codex (@codex) address that feedback".

[chatgpt-codex-connector]

Add the published Bedrock pricing

For this newly added model, the AWS Bedrock pricing page currently lists standard on-demand pricing for “NVIDIA Nemotron 3 Nano 30B A3B” as $0.06 input and $0.24 output per 1M tokens (checked https://aws.amazon.com/bedrock/pricing/). Leaving input_cost_per_mil_tokens and output_cost_per_mil_tokens out of this catalog entry makes getAvailableModels() expose the model as unpriced, so any cost estimates or accounting based on the static model metadata will be wrong for requests using nvidia.nemotron-nano-3-30b.

Useful? React with 👍 / 👎.