Skip to content

Add new AWS Bedrock Models 03-26#24652

Open
emerzon wants to merge 3 commits intoBerriAI:mainfrom
emerzon:bedrock_new_models_0326
Open

Add new AWS Bedrock Models 03-26#24652
emerzon wants to merge 3 commits intoBerriAI:mainfrom
emerzon:bedrock_new_models_0326

Conversation

@emerzon
Copy link
Contributor

@emerzon emerzon commented Mar 26, 2026

Adds support for AWS Bedrock:

  • writer.palmyra-vision-7b-v1
  • minimax.minimax-m2.5
  • nvidia.nemotron-super-3-120b
  • zai.glm-5

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

  • I have Added testing in the tests/test_litellm/ directory, Adding at least 1 test is a hard requirement - see details
  • My PR passes all unit tests on make test-unit
  • My PR's scope is as isolated as possible, it only solves 1 specific problem
  • I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Delays in PR merge?

If you're seeing a delay in your PR being merged, ping the LiteLLM Team on Slack (#pr-review).

CI (LiteLLM team)

CI status guideline:

  • 50-55 passing tests: main is stable with minor issues.
  • 45-49 passing tests: acceptable but needs attention
  • <= 40 passing tests: unstable; be careful with your merges and assess the risk.
  • Branch creation CI run
    Link:

  • CI run for the last commit
    Link:

  • Merge / cherry-pick CI run
    Links:

Type

🆕 New Feature

@vercel
Copy link

vercel bot commented Mar 26, 2026

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

Project Deployment Actions Updated (UTC)
litellm Ready Ready Preview, Comment Mar 27, 2026 1:49am

Request Review

@codspeed-hq
Copy link
Contributor

codspeed-hq bot commented Mar 26, 2026

Merging this PR will not alter performance

✅ 16 untouched benchmarks


Comparing emerzon:bedrock_new_models_0326 (aef8a48) with main (8f425ec)

Open in CodSpeed
@greptile-apps
Copy link
Contributor

greptile-apps bot commented Mar 26, 2026

Greptile Summary

This PR registers four new AWS Bedrock models — writer.palmyra-vision-7b, minimax.minimax-m2.5, nvidia.nemotron-super-3-120b, and zai.glm-5 — by adding entries to the two pricing/context-window JSON files. Each model receives a provider-level global entry (using bedrock_converse) plus per-region entries (using bedrock) covering APAC, EU, US, and SA regions with region-appropriate pricing tiers, including flex and priority cost multipliers.

  • writer.palmyra-vision-7b is added as a vision-capable bedrock_converse model; no per-region breakdown is included (unlike the other three new models), which is consistent with other non-cross-region Writer models in this file.
  • minimax.minimax-m2.5, nvidia.nemotron-super-3-120b, and zai.glm-5 are added across 11–13 AWS regions each with per-region pricing and supports_service_tier: true.
  • The confirmed intentional absence of zai.glm-5 from eu-central-1, eu-west-1, and eu-south-1 is reflected correctly.
  • Two minor, unrelated whitespace/formatting changes are made to existing supported_regions arrays in the same files.
  • No tests were added, which is listed as a hard requirement in the pre-submission checklist.

Confidence Score: 4/5

Safe to merge after confirming the correct AWS model ID for writer.palmyra-vision-7b; all other additions are well-structured and low-risk.

The PR is a pure JSON data change — no runtime logic is altered. The only actionable concern is whether writer.palmyra-vision-7b (the registered key) matches the actual AWS Bedrock model ID, since the PR description names it writer.palmyra-vision-7b-v1. If the key is wrong, that model's cost tracking and routing lookups would silently fall back to defaults for that one model; existing functionality is unaffected. All other additions are internally consistent, the regional coverage discrepancy for zai.glm-5 was confirmed intentional, and both JSON files are in sync.

model_prices_and_context_window.json and litellm/model_prices_and_context_window_backup.json — verify the writer.palmyra-vision-7b key name against the actual AWS model ID

Important Files Changed

Filename Overview
model_prices_and_context_window.json Adds global and per-region pricing entries for 4 new Bedrock models (writer.palmyra-vision-7b, minimax.minimax-m2.5, nvidia.nemotron-super-3-120b, zai.glm-5); includes trivial supported_regions formatting changes; potential key name mismatch for the writer model
litellm/model_prices_and_context_window_backup.json Mirrors main JSON additions exactly — same four new models added across same regions; inherits the same potential writer model key-name issue

Flowchart

%%{init: {'theme': 'neutral'}}%% flowchart TD User["User calls bedrock/model"] --> Lookup["litellm model_info lookup\nin model_prices_and_context_window.json"] Lookup --> RegionKey["Try bedrock/region/model_id\ne.g. bedrock/us-east-1/minimax.minimax-m2.5"] Lookup --> GlobalKey["Fallback: try model_id\ne.g. minimax.minimax-m2.5\n(bedrock_converse provider)"] RegionKey --> Hit1["Region-specific pricing\n+ supports_service_tier"] GlobalKey --> Hit2["Default pricing\n(us-east-1 rates used)"] RegionKey -- not found --> GlobalKey subgraph NewModels["New models added in this PR"] M1["writer.palmyra-vision-7b\nglobal only · bedrock_converse\nsupports_vision: true"] M2["minimax.minimax-m2.5\n11 regions + global"] M3["nvidia.nemotron-super-3-120b\n11 regions + global"] M4["zai.glm-5\n10 regions + global\n(intentionally absent from 3 EU regions)"] end 
Loading

Reviews (3): Last reviewed commit: "Fix Palmyra model ID" | Re-trigger Greptile

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

1 participant