All issues

Search results

/v1/models returns an empty list under load-on-demand, so OpenAI-compatible clients that discover models see none

Status: Open.

#130 In madroidmaq/mlx-omni-server;

· tonioloewald opened on Aug 6, 2026

MLX_VLM_ONLY_MODELS gates vision to a single architecture ('gemma4'); every other VLM falls through to mlx_lm and fails

Status: Open.

#129 In madroidmaq/mlx-omni-server;

· tonioloewald opened on Aug 6, 2026

chat/completions rejects the standard OpenAI image_url content block (Pydantic types content array values as str)

Status: Open.

#128 In madroidmaq/mlx-omni-server;

· tonioloewald opened on Aug 6, 2026

Tool calls in <function>NAME</function> + bare JSON {args} format not parsed

Status: Open.

#126 In madroidmaq/mlx-omni-server;

· ddavisqa opened on Jun 22, 2026

Your project is featured in Awesome MLX

Status: Open.

#118 In madroidmaq/mlx-omni-server;

· raullenchai opened on Mar 26, 2026

Tequila/Sherry "BitNet" support

Status: Open.

#104 In madroidmaq/mlx-omni-server;

· TomLucidor opened on Jan 27, 2026

Can this Server handle Linear Attention LLMs?

Status: Open.

#102 In madroidmaq/mlx-omni-server;

· TomLucidor opened on Jan 9, 2026

Suggest providing a default model path setting

Status: Open.

#101 In madroidmaq/mlx-omni-server;

· xrj-com opened on Jan 2, 2026

Caching (I think) causing the first token from a previous request to be prepended to new requests

Status: Open.

#80 In madroidmaq/mlx-omni-server;

· Madd0g opened on Oct 9, 2025

gpt-oss reasoning support

Status: Open.

#79 In madroidmaq/mlx-omni-server;

· kleincode opened on Oct 9, 2025

Could not extract base_model from model_name

Status: Open.

#78 In madroidmaq/mlx-omni-server;

· cadeon opened on Sep 30, 2025

deallocating memory error

Status: Open.

#75 In madroidmaq/mlx-omni-server;

· yz342 opened on Aug 26, 2025