mirror of
https://github.com/BerriAI/litellm.git
synced 2026-09-14 23:21:35 +00:00
e8311b586c
Add Friendli /serverless/v1/models as a data source to the weekly
auto_update_price_and_context_window_file.py sync.
For each Friendli model we emit a friendliai/{id} entry. When the
model's base_model already exists in litellm (under any provider
prefix), capability flags (supports_prompt_caching, supports_vision,
reasoning effort flags, etc.) are inherited from that curated entry;
Friendli-supplied pricing, context/output limits, modalities, and
reasoning options always override.
Reasoning option type=effort values (none/minimal/low/medium/high/
xhigh/max) are mapped to the matching litellm supports_*_reasoning_effort
boolean flags. budget_tokens with min=-1 means 'no budget limit' and
is left implicit -- litellm has no equivalent field.
6 new entries added (K-EXAONE-236B, MiniMax-M2.5, DeepSeek-V3.2,
GLM-5.1, gemma-4-31B-it, GLM-5.2); schema re-validated.
backup JSON intentionally untouched: ci_cd/check_files_match.py keeps
it in sync with main during CI, so it stays out of this diff.
|
||
|---|---|---|
| .. | ||
| _agent_shin_actions.py | ||
| agent_shin_shared.py | ||
| assert_ci_coverage.py | ||
| assert_workflow_dir_hygiene.py | ||
| auto_update_price_and_context_window_file.py | ||
| close_duplicate_issues.py | ||
| close_low_quality_prs.py | ||
| detect_changes.sh | ||
| run_llm_translation_tests.py | ||
| scan_keywords.py | ||
| select_ui_test_scope.sh | ||
| triage-requirements.txt | ||
| triage_with_llm.py | ||
| uv_sync_with_retries.sh | ||