26c9ce1288
* tests : fix fetch_server_test_models.py * server: to_json_oaicompat cached_tokens Adds OpenAI and Anthropic compatible information about the number of cached prompt tokens used in a response.