Skip to content

fix(cost-map): restore later azure Models API retirement dates and date gpt-6.1-sol - #44072

Merged
kerry-berri merged 2 commits into
mainfrom
litellm-providers/price-sync-azure
Oct 2, 2026
Merged

kerry-berri merged 2 commits into
mainfrom
litellm-providers/price-sync-azure

Conversation

@berriai-litellm-provider-info-sync

@berriai-litellm-provider-info-sync berriai-litellm-provider-info-sync Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

TLDR

Problem this solves:

How it solves it:

  • Restores the later Models API date on the eight rows and adds the gpt-6.1-sol date, deprecation_date only, merge mode keeps every other field
  • Same edit in model_prices_and_context_window.json and litellm/model_prices_and_context_window_backup.json

Rows

All rows are kind api, from the Azure Models API (the list source of GET /api/cli/list/azure): https://management.azure.com/subscriptions/c873328e-b572-4770-8dff-aaeb6f1f0e79/providers/Microsoft.CognitiveServices/locations/eastus2/models?api-version=2024-10-01, field model.deprecation.inference, fetched 2026-10-01. Schedule page: https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule (updated 2026-09-23), fetched raw

Row Before After Models API Schedule page cell
azure/gpt-4.1-nano 2026-10-14 2027-04-14 gpt-4.1-nano 2025-04-14: 2027-04-14T00:00:00Z gpt-4.1-nano | 2025-04-14 | Deprecated | 2026-10-14
azure/gpt-4.1-nano-2025-04-14 2026-10-14 2027-04-14 same same
azure/us/gpt-4.1-nano 2026-10-14 2027-04-14 same same
azure/us/gpt-4.1-nano-2025-04-14 2026-10-14 2027-04-14 same same
azure/eu/gpt-4.1-nano 2026-10-14 2027-04-14 same same
azure/gpt-4o-transcribe 2026-10-15 2026-12-31 gpt-4o-transcribe 2025-03-20: 2026-12-31T00:00:00Z gpt-4o-transcribe | 2025-03-20 | GA | 2026-10-15
azure/gpt-realtime-2.1 2027-06-25 2027-07-31 gpt-realtime-2.1 2026-07-07: 2027-07-31T00:00:00Z gpt-realtime-2.1 | 2026-07-07 | GA | 2027-06-25
azure/gpt-realtime-2.1-mini 2027-06-25 2027-07-31 gpt-realtime-2.1-mini 2026-07-07: 2027-07-31T00:00:00Z gpt-realtime-2.1-mini | 2026-07-07 | GA | 2027-06-25
azure/gpt-6.1-sol none 2028-03-11 gpt-6.1-sol 2026-09-29: 2028-03-11T00:00:00Z no row
azure/gpt-6.1-sol-2026-09-29 none 2028-03-11 same no row

The gpt-4.1-nano date is the case ruling azure-conflicting-dates-latest-wins was decided on. azure/gpt-4o-transcribe had no source, so it now points at the Models API URL; every other row keeps its existing source. azure_ai/gpt-6.1-sol stays undated like its sibling azure_ai/gpt-6-sol

Mapping

The list API emits OpenAI.<model>.<version> ids (for example OpenAI.gpt-4.1-nano.2025-04-14), mapped to azure/<model> and azure/<model>-<version>. That naming mismatch is why the dispatcher reported most azure/* OpenAI rows as delisted; they are still listed

Delisted by the provider

Absent from the eastus2 Models API under any name and left undated (no retirement date stated on the schedule or retired models pages): azure/chat-latest, azure/computer-use-preview, azure/container, azure/gpt-3.5-turbo, azure/gpt-3.5-turbo-instruct-0914, azure/gpt-35-turbo, azure/gpt-35-turbo-instruct, azure/gpt-35-turbo-instruct-0914, azure/gpt-4-0125-preview, azure/gpt-4-1106-preview, azure/gpt-4-turbo, azure/gpt-4-turbo-vision-preview, azure/gpt-5.2-pro, azure/gpt-5.5-pro, azure/gpt-5.6, azure/o1-mini, azure/o3-pro

Absent and already dated: azure/ada, azure/command-r-plus, azure/gpt-35-turbo-16k, azure/gpt-35-turbo-16k-0613, azure/gpt-4, azure/gpt-4-0613, azure/gpt-4-32k, azure/gpt-4-32k-0613, azure/gpt-4.5-preview, azure/gpt-5-chat, azure/gpt-5.1-chat, azure/gpt-5.2-chat, azure/gpt-5.3-chat, azure/o3-deep-research, azure/tts-1, azure/tts-1-hd, azure/whisper-1

azure/mistral-large-2402 and azure/mistral-large-latest stay undated per azure-mistral-large-conflicting-retirement-stay-undated. Coverage gap rather than delisting: azure/sora-2-pro, azure/sora-2-pro-high-res (video)

Changed pages

The Cohere pricing page matches the catalog (Command A $2.50/$10.00, Command A Plus $0.80/$3.20, embed-v4 $0.12, Rerank v4 Fast $2.00 and Pro $2.50 per 1K searches). The AOAI pricing page serves no static prices. The whats-new page has no entry after May 2026. The Claude models page only adds the gated Mythos preview, left out per azure-claude-mythos-leave-out-until-priced. models-sold-directly limits match the azure rows already reconciled in #43759

Pre-Submission checklist

  • I have added meaningful tests
  • My PR's scope is as isolated as possible; it only solves 1 specific problem

Note

Low Risk
Metadata-only cost-map JSON changes; no runtime logic, pricing fields, or auth paths are modified.

Overview
Updates Azure model deprecation_date metadata in model_prices_and_context_window.json and its backup so retirement reflects the later Azure Models API inference date when it disagrees with the public schedule—undoing a recent registry sync that had reverted eight rows to earlier schedule dates.

gpt-4.1-nano variants (global, US, EU) move from 2026-10-14 to 2027-04-14; gpt-realtime-2.1 and mini from 2027-06-25 to 2027-07-31; gpt-4o-transcribe from 2026-10-15 to 2026-12-31, with a new source pointing at the eastus2 Models API. gpt-6.1-sol and gpt-6.1-sol-2026-09-29 gain deprecation_date: 2028-03-11 where they were previously undated.

Reviewed by Cursor Bugbot for commit a490838. Bugbot is set up for automated code reviews on this repo. Configure here.

…te gpt-6.1-sol

Price-Sync: litellm-providers
@CLAassistant

CLAassistant commented Oct 1, 2026 •

Copy link
Copy Markdown

CLA assistant check
Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
0 out of 2 committers have signed the CLA.

❌ berriai-litellm-provider-info-sync[bot]
❌ kerry-berri
You have signed the CLA already but the status is still pending? Let us recheck it.

@greptile-apps

greptile-apps Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

RetriggerConfidence Score: 4/5

[Medium risk] Updates model deprecation dates in the pricing configuration.

The PR appears safe to merge, with non-blocking provenance and regression-test improvements recommended.

Findings

  1. P2 Source is not publicly verifiable ▶
  2. P2 Restored dates lack regression tests ▶

Summary

The PR restores later Azure retirement dates for eight catalog entries, adds dates for two gpt-6.1-sol entries, and keeps the primary and backup catalogs aligned.

  • The added transcription-model source is not publicly verifiable.
  • Value-specific tests would protect the restored dates from another registry rollback.

Reviews (1) · Last reviewed commit: "fix(cost-map): restore later azure Model..."

"max_output_tokens": 2000,
"mode": "audio_transcription",
"output_cost_per_token": 1e-05,
"source": "https://management.azure.com/subscriptions/c873328e-b572-4770-8dff-aaeb6f1f0e79/providers/Microsoft.CognitiveServices/locations/eastus2/models?api-version=2024-10-01",

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Source is not publicly verifiable The new source for azure/gpt-4o-transcribe is a subscription-scoped Azure management API URL, not a public provider model page. Readers without access to that subscription cannot use it to verify the revised retirement date. Please use a publicly verifiable source or make the provenance accessible.

Note: If this suggestion doesn't match your team's coding style, reply to this and let me know. I'll remember it for next time!

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

That context resolves my concern. The Azure Models API is the authoritative input for this registry sync, while the public retirement schedule documents the conflicting date and explains why the later API value wins. Since the PR body records the endpoint, field, region, and fetch date, the subscription-scoped URL is reasonable provenance for this metadata field. No change is needed.

},
"azure/gpt-4.1-nano": {
"deprecation_date": "2026-10-14",
"deprecation_date": "2027-04-14",

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

P2 Restored dates lack regression tests The updated Azure retirement dates have no value-specific assertions. The existing schema and catalog-consistency checks would still pass if a rolling registry update reverted these dates again. Please add assertions for the affected model entries so this fix remains protected.

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Understood. Given this is intentionally an isolated catalog-data sync, I won’t require individual-date regression tests in this PR. The existing schema/catalog checks still validate structure rather than these exact values, so a follow-up test or generated catalog validation would be useful to prevent a later registry sync from reverting the dates.

@cursor cursor Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

Reviewed by Cursor Bugbot for commit 2388307. Configure here.

@codspeed

codspeed Bot commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

Merging this PR will not alter performance

✅ 31 untouched benchmarks


Comparing litellm-providers/price-sync-azure (a490838) with main (d7e6154)

Open in CodSpeed

@codecov

codecov Bot commented Oct 1, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

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

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant