2026-03-09-azure-updates

[marty-sullivan]

Pre-Submission checklist

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

• 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

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

Changes

Adding new Azure OpenAI models to cost map:

• gpt-5.3-chat
• gpt-5.4
• gpt-5.4-pro

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
litellm	Error		Mar 9, 2026 3:33pm

[marty-sullivan]

@greptileai please review

[greptile-apps]

Greptile Summary

This PR adds five new Azure OpenAI model entries to the LiteLLM cost map in both model_prices_and_context_window.json and the backup file. The three model families added are gpt-5.3-chat, gpt-5.4, and gpt-5.4-pro, following the existing conventions for Azure model pricing metadata.

Key observations:

• All previously flagged review issues have been resolved: undated aliases, supports_service_tier, and priority cost fields are all correctly present.
• azure/gpt-5.3-chat is added as an undated-only entry (no dated version), which is consistent with the existing azure/gpt-5.3-codex pattern for the 5.3 generation.
• azure/gpt-5.4 (undated + 2026-03-05) has full priority pricing (_priority and _above_272k_tokens_priority variants), supports_service_tier: true, and mode: "chat".
• azure/gpt-5.4-pro (undated + 2026-03-05) uses mode: "responses", has no priority pricing fields, and correctly omits supports_service_tier — consistent with the azure/gpt-5.2-pro pattern.
• Both JSON files (model_prices_and_context_window.json and litellm/model_prices_and_context_window_backup.json) are kept in sync with identical additions.

Confidence Score: 5/5

• This PR is safe to merge — it only adds JSON pricing metadata entries with no runtime code changes.
• All changes are confined to JSON configuration files. All previously raised review concerns (missing undated aliases, missing supports_service_tier, missing priority cost fields) have been fully addressed. The new entries are internally consistent and follow established conventions for their respective model generations. No logic, tests, or code paths are affected.
• No files require special attention.

Important Files Changed

Filename	Overview
model_prices_and_context_window.json	Adds 5 new Azure OpenAI model entries (azure/gpt-5.3-chat, azure/gpt-5.4, azure/gpt-5.4-2026-03-05, azure/gpt-5.4-pro, azure/gpt-5.4-pro-2026-03-05) with all required pricing, context window, capability, and service-tier fields. All previously flagged issues (undated aliases, supports_service_tier, and priority cost fields) have been resolved. Entries are internally consistent.
litellm/model_prices_and_context_window_backup.json	Backup JSON kept in sync with the root model_prices_and_context_window.json — identical new entries added; no issues found.

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A[New Azure Models - This PR] --> B["azure/gpt-5.3-chat\n(undated only)\nmode: chat | 128K ctx\nhas priority pricing\nsupports_service_tier: true"]
    A --> C["azure/gpt-5.4\n(undated alias)\nmode: chat | 1050K ctx\nhas priority pricing\nsupports_service_tier: true"]
    A --> D["azure/gpt-5.4-2026-03-05\n(dated)\nmode: chat | 1050K ctx\nhas priority pricing\nsupports_service_tier: true"]
    A --> E["azure/gpt-5.4-pro\n(undated alias)\nmode: responses | 1050K ctx\nno priority pricing\nno service_tier"]
    A --> F["azure/gpt-5.4-pro-2026-03-05\n(dated)\nmode: responses | 1050K ctx\nno priority pricing\nno service_tier"]
    C <--> |mirrors| D
    E <--> |mirrors| F
    B --- G["Pattern: consistent with\nazure/gpt-5.3-codex\n(undated only)"]
    C --- H["Pattern: consistent with\nazure/gpt-5.2 + azure/gpt-5.2-2025-12-11\n(undated + dated pair)"]

Loading

_{Last reviewed commit: ed11982}