feat(azure_ai): add Grok 4.20 model metadata

[emerzon]

Summary

• add Azure AI Foundry pricing metadata for azure_ai/grok-4-20-reasoning and azure_ai/grok-4-20-non-reasoning
• include Grok 4.20 Azure capability metadata for multimodal input, vision, tool calling, and the reasoning split
• add focused Azure AI tests for model info and cost calculation

Pricing / sources

• Azure announcement: https://techcommunity.microsoft.com/blog/azure-ai-foundry-blog/grok-4-20-is-now-available-in-microsoft-foundry/4509377
• Azure catalog pages for grok-4-20-reasoning and grok-4-20-non-reasoning
• pricing used: $2.00 / 1M input tokens, $6.00 / 1M output tokens

Testing

• conda run -n litellm pytest tests/test_litellm/llms/azure_ai/test_azure_ai_cost_calculator.py tests/test_litellm/llms/azure_ai/test_azure_ai_grok_420_metadata.py -q

[codspeed-hq]

Merging this PR will not alter performance

✅ 16 untouched benchmarks

---

_{Comparing emerzon:egomes/azure-grok-4-20 (8b48cb0) with main (934ecdc)}

[greptile-apps]

Greptile Summary

This PR adds model metadata entries for azure_ai/grok-4-20-reasoning and azure_ai/grok-4-20-non-reasoning to both JSON price files, along with a focused mock-only test suite. The entries are well-formed, pricing ($2/$6 per M tokens) matches the referenced Azure announcement, capability flags are consistent with existing azure_ai grok siblings, and the previously flagged supports_response_schema omission has been addressed in the latest commit.

Confidence Score: 5/5

Safe to merge — pure metadata addition with no logic changes and full test coverage.

All three files are clean: JSON entries are well-formed and in sync, pricing and capability flags are consistent with existing azure_ai grok entries, and the new tests are mock-only with no real network calls. No P0 or P1 findings.

No files require special attention.

Important Files Changed

Filename	Overview
model_prices_and_context_window.json	Adds two new azure_ai/grok-4-20 model entries with correct pricing ($2/6 per M tokens), context window (256k), capability flags including supports_response_schema, supports_vision, supports_reasoning (reasoning variant only), consistent with existing azure_ai grok sibling entries.
litellm/model_prices_and_context_window_backup.json	Mirrors the same two grok-4-20 entries from the primary JSON; both files are in sync including the supports_response_schema: true field.
tests/test_litellm/llms/azure_ai/test_azure_ai_grok_420_metadata.py	New test file with three well-structured mock-only test functions covering model info, raw cost entry, and cost-per-token calculation; correctly uses the backup JSON via LITELLM_LOCAL_MODEL_COST_MAP with no real network calls, complying with the mock-only test rule.

_{Reviews (2): Last reviewed commit: "Add response schema support for Azure AI..." | Re-trigger Greptile}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[Copilot]

Pull request overview

Adds Azure AI Foundry metadata and pricing for the new Grok 4.20 model variants, and validates the metadata + cost calculation behavior via targeted tests.

Changes:

• Added pricing/context window entries for azure_ai/grok-4-20-reasoning and azure_ai/grok-4-20-non-reasoning (including multimodal + capability flags).
• Updated the packaged backup cost map to include the same Grok 4.20 entries.
• Introduced focused pytest coverage for get_model_info() metadata and Azure AI cost_per_token() calculations for Grok 4.20.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated no comments.

File	Description
tests/test_litellm/llms/azure_ai/test_azure_ai_grok_420_metadata.py	New tests asserting Grok 4.20 Azure AI model metadata and verifying cost calculations against the added pricing.
model_prices_and_context_window.json	Adds Grok 4.20 Azure AI entries to the main (repo) model cost map.
litellm/model_prices_and_context_window_backup.json	Adds Grok 4.20 Azure AI entries to the bundled local backup cost map used for local/forced pricing lookups and tests.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[github-actions]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.