All issues

[Bug, Issue]: ASR model (whisper-large-v3-turbo) not explicitly named in v1/models, so doesn't work with stricter openAI compatible connections

#746

· Seyij opened

on Sep 19, 2026

[Feature]: Add support for Cactus Compute Needle model

#745

· julianxhokaxhiu opened

on Sep 18, 2026

No valid checkpoint to restore + Max length reached on multi-turn tool calls (Qwen3.6-MoE, NPU)

#744

· winaes13-ux opened

on Sep 18, 2026

Truncated tool call is emitted as a complete one (unterminated heredoc reaches the client)

#741

· Javinator9889 opened

on Sep 18, 2026

Non-streaming responses leak <think> tags into content for qwen3.5, qwen3.6-moe, qwen3vl and qwen3.5-omni

#739

· Javinator9889 opened

on Sep 18, 2026

[Feature]: KV-cache reuse for shared-prefix one-shot requests (prefix caching / partial truncate)

#737

· thunderbug1 opened

on Sep 17, 2026

Prompt cache: identical resend never hits, partial match discards everything, and nothing is sent during prefill

#733

· Javinator9889 opened

on Sep 17, 2026

v1.0.4 regression: embed-gemma:300m embeddings via --embed 1 are wrong (uncorrelated with v0.9.39 on identical weights)

#732

· cbei1261-gmail-com opened

on Sep 17, 2026

[Feature/Model Request]: Support for NPU-accelerated Text-to-Speech (TTS) models

#731

· Seyij opened

on Sep 17, 2026

[Issue]: `usage.prompt_tokens` on the OpenAI-compatible endpoints reports only the tokens prefilled on the current turn

#730

· Javinator9889 opened

on Sep 17, 2026

[Documentation]: How to generate SRT/VTT transcripts

#726

· vincent-sparks opened

on Sep 13, 2026

[Issue]: [Bug][Linux] flm serve crash: C++ assertion failure '!this->empty()' in std::vector<logits_t>::back()

#725

· austrasien opened

on Sep 12, 2026