fix(schema): classify off_peak_pricing as a structured object in the model prices schema generator - #41847

TLDR

Problem this solves:

• The schema generator rejects off_peak_pricing as an unclassified key
• The provider price sync could not publish the openrouter catalog because of it

How it solves it:

• Adds off_peak_pricing to OBJECT_KEYS as a structured object schema
• Shape mirrors the runtime OffPeakPricing TypedDict: hours_utc, windows, weekday_timezone, five override rates
• Windows must be HH:MM-HH:MM, weekday names must be ones the runtime recognizes, and a block needs hours_utc or windows
• Regression tests accept a valid block and reject malformed ones

User Flow

Before: the price sync bot cannot publish an openrouter catalog update that carries an off-peak rate block

1. The sync bot fetches openrouter prices and finds a model with a time-of-day discount
2. It tries to open a catalog PR against BerriAI/litellm and reports in Slack: "The catalog uses fields its schema does not define and the sync cannot classify: off_peak_pricing"
3. Running python ci_cd/generate_model_prices_schema.py on that catalog prints "Unclassified keys in model_prices_and_context_window.json: off_peak_pricing" and no PR is opened

After: the same catalog validates and the sync can publish

1. The sync bot fetches openrouter prices and finds a model with a time-of-day discount
2. python ci_cd/generate_model_prices_schema.py renders off_peak_pricing as an object with hours_utc, windows, weekday_timezone and the override rates, and the catalog validates against the result
3. The sync opens its catalog PR as usual

Relevant issues

Companion change in BerriAI/litellm-providers keeps the sync side's schema classification identical to this generator

Affected release

Linear ticket

Resolves LIT-8128

Pre-Submission checklist

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

• I have added meaningful tests
• The handful of test files covering my change pass locally, e.g. uv run pytest tests/test_litellm/<your_test_file>.py -v. Leave the suites (make test-unit-*, make test-unit) to CI: it finishes in ~15 minutes where a laptop takes an hour or more
• My PR passes all required CI/CD checks (e.g., lint, schema.d.ts sync check, etc.)
• 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)

Delays in PR merge?

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

Screenshots / Proof of Fix

This change only touches the schema generator that CI and the price sync run, so the proof is the generator itself, run in a worktree at each commit against the current catalog plus one entry carrying an off-peak block:

"openrouter/deepseek/deepseek-chat-off-peak-demo": {
    "litellm_provider": "openrouter",
    "mode": "chat",
    "input_cost_per_token": 2e-06,
    "output_cost_per_token": 8e-06,
    "off_peak_pricing": {"hours_utc": "16:30-00:30", "input_cost_per_token": 1e-06, "output_cost_per_token": 4e-06}
}

Before (db756b9)

1. python ci_cd/generate_model_prices_schema.py --check
2. Output: Unclassified keys in model_prices_and_context_window.json: off_peak_pricing. Add them to the key tables in generate_model_prices_schema.py and rerun it.

After (4b215e2)

1. python ci_cd/generate_model_prices_schema.py --check
2. Output: model_prices_and_context_window.schema.json is out of sync with model_prices_and_context_window.json. Run python ci_cd/generate_model_prices_schema.py and commit the result. (the key is classified; the committed schema simply does not yet carry it because no entry on main uses it)
3. python ci_cd/generate_model_prices_schema.py then --check again: model_prices_and_context_window.schema.json is in sync and model_prices_and_context_window.json validates against it

On the unmodified main catalog --check still passes at the PR tip, so the committed schema is unchanged by this PR

Type

🐛 Bug Fix

Caveats (if any)

Low

• The committed schema file does not change until a catalog entry actually uses off_peak_pricing
• The generator has preexisting ruff findings (UP045, T201) that this PR does not touch

Final Attestation

• The tests check the right things, including the edge cases, and regressions in the respective real-world customer use-cases are not possible after this PR

Link to Devin session: https://app.devin.ai/sessions/b3650eb76464401f9bd6e0e7264bf62b
Open in Devin Desktop: https://app.devin.ai/desktop/session/b3650eb76464401f9bd6e0e7264bf62b?variant=devin