Request New Models/Provider/Endpoints - Jan 2026

[ishaan-jaff]

Request New Models, Providers & Endpoints - Jan 2026

This is the central place for the LiteLLM community to request and vote on new integrations.

How to Request

Before posting, search existing requests to avoid duplicates.

To request a new integration:

1. Comment below with:

   • Provider/Model name
   • Documentation link (if available)
   • Your use case

2. Upvote existing requests you want to see added

Prioritization

Requests are prioritized based on:

• Community upvotes

Already Supported

Check our providers documentation - we currently support 100+ models across OpenAI, Anthropic, Azure, AWS Bedrock, Vertex AI, and more.

---

For urgent production needs, contact us directly at support@berri.ai

[ishaan-jaff]

• /responses/compression API
• docs https://platform.openai.com/docs/api-reference/responses/compact

Codex CLI now uses this endpoint

[b3nw]

Provider/Model name: Firmware.ai
Documentation link (if available): https://docs.firmware.ai/
Your use case: leverage subscription

#19547 submitted

[cgilly2fast]

Lets go! thank you for this Ben!

[kaaduu]

Yeah we want it!

[xapids]

Please lets get this subscription supported 🙏

[robinr2]

Best provider!

[ryo9999]

Provider/Model name: zenmux.ai
Documentation link (if available): https://zenmux.ai/docs/about/intro.html
Your use case: leverage subscription

[jweisman11]

Provider/Model name: Intel OpenVINO (specifically supporting OpenVINO Model Server or ovms)
Documentation link (if available): https://github.com/openvinotoolkit/openvino and https://docs.openvino.ai/2025/index.html, but specifically https://docs.openvino.ai/2025/model-server/ovms_docs_clients_genai.html

Your use case:

Our team is migrating to Small Language Models (SLMs) hosted on Intel Xeon-based EC2 instances (M7i/z1d families). We use the OpenVINO framework to compress these models for high-performance CPU inference, achieving low latency without the need for expensive GPUs.

Currently, we use the LiteLLM Proxy for all our LLM routing, cost tracking, and unified API access. However, because LiteLLM lacks a native OpenVINO adapter, we cannot easily route traffic to our self-hosted OpenVINO endpoints through the same proxy.

Why this is valuable:

• Cost Efficiency: Enables teams to use Intel CPU-based instances for production LLMs, significantly reducing cloud spend compared to GPU instances.
• Edge/On-Prem Deployment: Many enterprise environments run on Intel hardware; adding OpenVINO support makes LiteLLM the go-to proxy for local/edge AI deployments.
• Standardization: Allows us to manage OpenVINO-hosted SLMs alongside OpenAI/Anthropic/Bedrock models using a single LiteLLM configuration.

[kilaka2]

Request support for xAI models: grok-4.20-multi-agent-beta-0309, grok-4.20-beta-0309-reasoning, grok-4.20-beta-0309-non-reasoning

Provider/Model name: xAI / grok-4.20-multi-agent-beta-0309, xAI / grok-4.20-beta-0309-reasoning, xAI / grok-4.20-beta-0309-non-reasoning
Documentation link: https://docs.x.ai/developers/models
Use case: Our application currently supports grok-4-fast models, and we plan to extend it with the latest xAI models
Endpoint(s) needed: /responses