Search results

mlx_vlm.chat_ui with DiffusionGemma raises "There is no Stream(gpu, 1) in current thread." on MacBook Apple M1

Status: Open.

#1362 In Blaizzy/mlx-vlm;

· johnkea opened on Jun 12, 2026

DiffusionGemma long-context prefill: the masks are eating all the RAM

codex

needs-design

priority/p2

type/feature

Status: Open.

#1355 In Blaizzy/mlx-vlm;

· johntdavies opened on Jun 11, 2026

Diffusion path strips all special tokens at the token-id level, deleting the tool-call/channel markers its own gemma4 parser needs (tool_calls: null, 'thought' leaks into content)

codex

confirmed

priority/p1

type/bug

Status: Open.

#1351 In Blaizzy/mlx-vlm;

· dudeoverhere opened on Jun 11, 2026

Server batching: [concatenate] No arrays provided for concatenation in get_rope_index (Qwen3.5/3.6) with mixed-length vision requests — one failure takes down all in-flight requests

codex

needs-repro

priority/p1

type/bug

Status: Open.

#1346 In Blaizzy/mlx-vlm;

· gruenewald opened on Jun 10, 2026

gemma-4-e4b-it-4bit fails to load: load_model() drops strict kwarg before model.load_weights()

codex

confirmed

priority/p1

type/bug

Status: Open.

#1340 In Blaizzy/mlx-vlm;

· gierran opened on Jun 9, 2026

0.6.2 newest main branch source, APC enabled but not APC effects [Qwen 3.6 35B A3B mxfp4 with 5bit draft MTP model]

needs-repro

priority/p3

questionFurther information is requestedFurther information is requested

Status: Open.

#1339 In Blaizzy/mlx-vlm;

· NeoInBJ opened on Jun 9, 2026

vlm-mlx v0.6.2 crushed when switching model in request payload

needs-repro

priority/p1

type/bug

Status: Open.

#1335 In Blaizzy/mlx-vlm;

· tyh-carl opened on Jun 9, 2026

KV cache quantization silently skipped for QAT models — but QAT weights and KV cache are independent concerns

needs-repro

priority/p2

type/bug

Status: Open.

#1333 In Blaizzy/mlx-vlm;

· bonyohana opened on Jun 8, 2026

Chandra OCR model crashed on mlx-vlm 0.6.2

needs-repro

priority/p1

type/bug

Status: Open.

#1329 In Blaizzy/mlx-vlm;

· tyh-carl opened on Jun 8, 2026

How good are local models at recognising cats from a very small image?

priority/p3

questionFurther information is requestedFurther information is requested

Status: Open.

#1327 In Blaizzy/mlx-vlm;

· jrp2014 opened on Jun 7, 2026

0.6.x even installed from newest main branch source code, pure text RAG with qwen 3.6 35B A3B mxfp4 still could not using APC full speed in prefill

needs-repro

priority/p2

type/bug

Status: Open.

#1320 In Blaizzy/mlx-vlm;

· NeoInBJ opened on Jun 7, 2026

MTP speculative decoding crashes on Qwen3.6-35B-A3B

confirmed

priority/p1

type/bug

Status: Open.

#1317 In Blaizzy/mlx-vlm;

· huanglibo389 opened on Jun 6, 2026