Search results

CPU backend aborts on DeepSeek V4 PRO: grouped Q8_0 tensor has an unexpected layout

Status: Open.

#510 In antirez/ds4;

· mrighini opened on Jul 6, 2026

Non-streaming /v1/chat/completions leaks thinking into content when truncated (finish_reason: length) before </think>

Status: Open.

#509 In antirez/ds4;

· stefanwichmann opened on Jul 6, 2026

GLM 5.2 - Distributed Inference Fail

Status: Open.

#505 In antirez/ds4;

· fcrisafulli opened on Jul 5, 2026

CUDA ssd-streaming: expert cache thrashes on mixed-quant GGUFs (decode ~0.9 t/s); prefill drops the whole cache every request

Status: Open.

#503 In antirez/ds4;

· iCreil opened on Jul 5, 2026

PRO Q4 on two M3 Ultra 512GB: decode ~3.6 s/token on BOTH halves (~0.14 t/s); a single decode token is slower than a 512-token prefill

Status: Open.

#501 In antirez/ds4;

· treemanz opened on Jul 5, 2026

Many prefills while using Claude Code extension

Status: Open.

#500 In antirez/ds4;

· liqili opened on Jul 5, 2026

GLM 5.2 error: ds4: required tensor is missing: token_embd.weight

Status: Open.

#496 In antirez/ds4;

· links486 opened on Jul 4, 2026

What blocks --mtp together with --ssd-streaming? (data: it would matter most on disk-bound rigs)

Status: Open.

#495 In antirez/ds4;

· giannisanni opened on Jul 4, 2026

Add optional expert bundle sidecar for contiguous MoE expert reads

Status: Open.

#491 In antirez/ds4;

· GiorgioOppo opened on Jul 3, 2026

will be possible to run DS4 with Intel Arc B70 Gpu's ?

Status: Open.

#486 In antirez/ds4;

· arisaliti opened on Jul 2, 2026

Feature proposal: Decision gates for supervised agent / tool execution

Status: Open.

#485 In antirez/ds4;

· sdellava opened on Jul 2, 2026

CUDA + discrete GPU: page-cache residency when RAM ≥ model — measured before/after (decode +52%)

Status: Open.

#477 In antirez/ds4;

· fioruccione opened on Jun 29, 2026