feat: add LiteLLM as AI gateway provider

[RheagalFire]

Description

Adds LiteLLM as a new REST backend for spacy-llm, giving users access to 100+ LLM providers (OpenAI, Anthropic, Azure, Bedrock, Vertex AI, Groq, Ollama, etc.) through the LiteLLM Python SDK.

LiteLLM is added as a new provider under spacy_llm/models/rest/litellm/, following the exact same pattern as the existing OpenAI, Anthropic, and Cohere providers. It registers as spacy.LiteLLM.v1 in the model registry.

Files added:

• spacy_llm/models/rest/litellm/model.py -- LiteLLM class extending REST, uses litellm.completion() with drop_params=True
• spacy_llm/models/rest/litellm/registry.py -- @registry.llm_models("spacy.LiteLLM.v1") registration
• spacy_llm/models/rest/litellm/__init__.py -- exports

Files modified:

• spacy_llm/models/rest/__init__.py -- registered litellm module

Live E2E (Anthropic Claude Sonnet via Azure Foundry):

>>> litellm.completion(model='anthropic/claude-sonnet-4-6', messages=[...], drop_params=True)
Response: 4
Model: claude-sonnet-4-6

Example usage:

Config-based (spacy pipeline):

# config.cfg
[components.llm.model]
@llm_models = "spacy.LiteLLM.v1"
name = "anthropic/claude-haiku-4-5"

Python:

from spacy_llm.models.rest.litellm import LiteLLM

llm = LiteLLM(
    name="anthropic/claude-haiku-4-5",
    endpoint="",
    config={"temperature": 0},
    strict=True,
    max_tries=3,
    interval=1.0,
    max_request_time=30,
    context_length=None,
)
# Provider API keys read from env vars (e.g. ANTHROPIC_API_KEY)
responses = llm([["Summarize this text: ..."]])

Model names use the litellm provider/model-name format (e.g. anthropic/claude-haiku-4-5, openai/gpt-4o, bedrock/anthropic.claude-3-haiku).

Corresponding documentation PR

N/A

Types of change

New feature -- adds a new LLM provider backend.

Checklist

• I confirm that I have the right to submit this contribution under the project's MIT license.
• I ran all tests in tests and usage_examples/tests, and all new and existing tests passed. This includes
  • all external tests (i. e. pytest ran with --external)
  • all tests requiring a GPU (i. e. pytest ran with --gpu)
• My changes don't require a change to the documentation, or if they do, I've added all required information.

[RheagalFire]

cc @svlandeg @rmitsch

[svlandeg]

Hi @RheagalFire, I'm not involved in maintaining this library anymore, so please refrain from further pings 🙏