Search results

OpenVINO Model Server fails to load LFM2-24B-A2B-int4-ov

bugSomething isn't workingSomething isn't working

Status: Open.

#4405 In openvinotoolkit/model_server;

· ss4workonly opened on Jul 24, 2026

Enable_prefix_caching: Suspected cache inaccuracy during prefix reuse in Linear Attention Models

bugSomething isn't workingSomething isn't working

Status: Open.

#4376 In openvinotoolkit/model_server;

· wjlingz opened on Jul 15, 2026

Baremetal ovms_ubuntu24 binary fails on newer distros — libxml2.so.2 (+ ICU 74) is a runtime dep but not bundled

Status: Open.

#4362 In openvinotoolkit/model_server;

· exzile opened on Jul 9, 2026

Windows Manager: Updates page and staged package upgrade workflow

Status: Open.

#4351 In openvinotoolkit/model_server;

· exzile opened on Jul 1, 2026

[RFC] Standalone Windows installer + tray Manager for OVMS (install / repair / uninstall / update)

Status: Open.

#4347 In openvinotoolkit/model_server;

· exzile opened on Jun 30, 2026

Qwen3.5-35B-A3B-int4-ov gibberish output during high memory pressure

bugSomething isn't workingSomething isn't working

Status: Open.

#4345 In openvinotoolkit/model_server;

· TayQiXiang opened on Jun 30, 2026

Windows build fails on Visual Studio 2026 / newer MSVC toolsets (hardcoded VS2022 + toolset)

Status: Open.

#4333 In openvinotoolkit/model_server;

· exzile opened on Jun 27, 2026

/chat/completions fails ("Chat template not loaded correctly") for VL-derived / omni text decoders — Python-Jinja path leaves template null; MINJA mode would fix it but is not in the 2026.2.x release

Status: Open.

#4322 In openvinotoolkit/model_server;

· exzile opened on Jun 24, 2026

CPU_RUNTIME_CACHE_CAPACITY is unsupported in OVMS plugin_config but works in raw OpenVINO Runtime

Status: Open.

#4299 In openvinotoolkit/model_server;

· Isochronism opened on Jun 19, 2026

Gemma-4-26b-a4b-it-int4-ov gibberish output

bugSomething isn't workingSomething isn't working

Status: Open.

#4298 In openvinotoolkit/model_server;

· wjlingz opened on Jun 19, 2026

Feature request: CPU offload / HETERO:GPU,CPU support for models exceeding VRAM in continuous batching

Status: Open.

#4263 In openvinotoolkit/model_server;

· devedse opened on May 29, 2026

generation_config.json path override per LLM node

enhancementNew feature or requestNew feature or request

Status: Open.

#4233 In openvinotoolkit/model_server;

· korund opened on May 22, 2026