fix(pricing): apply OpenAI's gpt-5.6 terra/luna cut to Azure cost map

[mubashir1osmani]

TLDR

Problem this solves:

• Azure gpt-5.6 Terra/Luna cost map still used pre-cut rates
• Spend tracking over-billed Azure Terra/Luna after the 2026-07-30 cut
• openai and bedrock_mantle already matched the new rates

How it solves it:

• Lower Azure global Terra to $2/$12 per 1M tokens
• Lower Azure global Luna to $0.20/$1.20 per 1M tokens
• Scale us/eu data-zone rows at the usual 1.1x uplift
• Leave Sol unchanged; scale cache-read, priority, long-context with existing multipliers

source: https://azure.microsoft.com/en-us/blog/gpt-5-6-now-available-in-microsoft-foundry/?msockid=3fb522c6c76b6c1d328d3541c6886d42

Relevant issues

Linear ticket

Pre-Submission checklist

• I have added meaningful tests
• My PR passes all CI/CD checks (e.g., lint, format, unit tests)
• My PR's scope is as isolated as possible; it only solves 1 specific problem
• I have received a Greptile Confidence Score of at least 4/5 before requesting a maintainer review (Greptile reviews automatically once the PR is opened; only comment @greptileai to re-request a review after pushing changes)

Type

🐛 Bug Fix

Changes

Align azure/gpt-5.6-terra, azure/gpt-5.6-luna, and the matching azure/us/* and azure/eu/* data-zone rows with OpenAI's 2026-07-30 price cut (Terra -20%, Luna -80%). Sol is unchanged

Sources checked before the change:

• OpenAI pricing: https://developers.openai.com/api/docs/pricing (Terra $2/$12, Luna $0.20/$1.20 short context)
• OpenAI announcement: https://openai.com/index/advancing-the-price-performance-frontier-with-gpt-5-6/
• Existing LiteLLM openai and bedrock_mantle gpt-5.6 entries already at the new rates
• Azure global historically mirrors OpenAI global for these models; us/eu keep the 10% data-zone uplift

Unit coverage: test_generic_cost_per_token_azure_gpt56 asserts the new Azure global and eu data-zone input/output/cache-read rates

Screenshots / Proof of Fix

LITELLM_LOCAL_MODEL_COST_MAP=True python3 -m pytest \
  tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py::test_generic_cost_per_token_azure_gpt56 -q

.......                                                                  [100%]
7 passed in 0.70s

Cost-map check after the change (per-token rates):

python3 - <<'PY'
import json
d=json.load(open("model_prices_and_context_window.json"))
for k in [
  "azure/gpt-5.6-sol","azure/gpt-5.6-terra","azure/gpt-5.6-luna",
  "azure/eu/gpt-5.6-terra","azure/eu/gpt-5.6-luna",
  "gpt-5.6-terra","gpt-5.6-luna",
]:
  m=d[k]
  print(f"{k}: in={m['input_cost_per_token']} out={m['output_cost_per_token']} cache_r={m['cache_read_input_token_cost']}")
PY

azure/gpt-5.6-sol: in=5e-06 out=3e-05 cache_r=5e-07
azure/gpt-5.6-terra: in=2e-06 out=1.2e-05 cache_r=2e-07
azure/gpt-5.6-luna: in=2e-07 out=1.2e-06 cache_r=2e-08
azure/eu/gpt-5.6-terra: in=2.2e-06 out=1.32e-05 cache_r=2.2e-07
azure/eu/gpt-5.6-luna: in=2.2e-07 out=1.32e-06 cache_r=2.2e-08
gpt-5.6-terra: in=2e-06 out=1.2e-05 cache_r=2e-07
gpt-5.6-luna: in=2e-07 out=1.2e-06 cache_r=2e-08

[greptile-apps]

Greptile Summary

The PR updates Azure GPT-5.6 Terra and Luna rates to reflect the announced price reduction.

• Updates global and US/EU data-zone input, output, cache-read, priority, and long-context rates in both pricing maps.
• Updates cost-calculation expectations for the affected global and EU model variants.

Confidence Score: 5/5

The PR appears safe to merge.

No blocking failure remains, and both previously reported workflow issues are absent from the current HEAD.

Important Files Changed

Filename	Overview
model_prices_and_context_window.json	Updates primary Azure GPT-5.6 Terra/Luna pricing entries and regional data-zone rates.
litellm/model_prices_and_context_window_backup.json	Keeps the backup pricing map synchronized with the primary model pricing map.
tests/test_litellm/litellm_core_utils/llm_cost_calc/test_llm_cost_calc_utils.py	Updates Azure GPT-5.6 cost expectations to match the revised pricing metadata.

_{Reviews (3): Last reviewed commit: "chore: drop fork-only GHCR publish workf..." | Re-trigger Greptile}

[veria-ai]

PR overview

All previously flagged issues have been addressed. No open security concerns remain on this pull request.

Security review

No open security issues remain on this pull request.

Fixed/addressed: 2 · PR risk: 0/10

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[mubashir1osmani]

Closing: Azure has not published a GPT-5.6 Terra/Luna price cut. OpenAI cut rates on 2026-07-30, but Azure public pricing and Microsoft Q&A still reflect the pre-cut numbers. Reverting so we do not underbill Azure spend until Azure documents new rates.

[codspeed-hq]

Merging this PR will not alter performance

✅ 31 untouched benchmarks

---

_{Comparing mubashir1osmani:litellm_azure_gpt56_price_cut (0de901a) with litellm_internal_staging (e64536c)¹}

Footnotes

1. No successful run was found on litellm_internal_staging (e2950a8) during the generation of this report, so c93a336 was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

[mateo-berri]

@greptileai

[mateo-berri]

LGTM. Thanks!

[mihidumh]

@mateo-berri Azure has not updated their pricing yet, this should be reverted for now.