We have 2 ways of inference with hf models: 1. Transformers backend 2. vllm backend It is worth adding the inference via API for models from like openai or anthropic.