Skip to content

Pull requests: Blaizzy/mlx-vlm

New pull request New
59 Open 532 Closed
59 Open 532 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Reviews
Filter by reviews
No reviews Review required Approved review Changes requested
Assignee
Filter by who’s assigned
Assigned to nobody Loading
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Pull requests list

Improve text-only decode performance
#1105 opened May 4, 2026 by lucasnewman Collaborator Loading…
granite4_vision: support standard granite MLP backbone for 4.1-4b
#1104 opened May 3, 2026 by EliSchwartz Loading…
[WIP] Add APC prompt caching with disk persistence
#1103 opened May 3, 2026 by Blaizzy Owner • Draft
Fix: model discovery for single-file safetensors models & improve /v1/models reporting
#1101 opened May 2, 2026 by frandariem Loading…
Fix Gemma4 chat templating for Python generate()
#1099 opened May 1, 2026 by Girish011 Loading…
minicpmo / fastvlm: fix pixel cast on quantized language models
#1098 opened May 1, 2026 by contrapuntal Loading…
3 tasks done
1
feat(server): support tools / function-calling on /v1/responses
#1097 opened May 1, 2026 by jasonvassallo Loading…
fix: Batch generation breaks top-p sampling
#1094 opened Apr 30, 2026 by spicyneuron Contributor Loading…
fix: Prevent batched cache metadata lazy graph buildup
#1093 opened Apr 30, 2026 by spicyneuron Contributor Loading…
server: added loaded model's context size and tool call parser to /health endpoint
#1092 opened Apr 30, 2026 by goniz Contributor Loading…
1
Add Sapiens2 + RTMDet (top-down pose pipeline)
#1081 opened Apr 26, 2026 by Blaizzy Owner Loading…
8 tasks done
fix: remap per-layer quantization keys for mixed-bit models
#1078 opened Apr 26, 2026 by ivanfioravanti Contributor Loading…
2
Fix Unicode byte-fallback decoding in server streaming responses
#1076 opened Apr 26, 2026 by spicyneuron Contributor Loading…
fix(qwen3_5_moe): route nested model.language_model.visual.* to vision_tower (#1057)
#1075 opened Apr 26, 2026 by machiabeli Loading…
3 tasks done
1
Fix stale test_utils.py regressions + extract get_class_predicate
#1071 opened Apr 25, 2026 by mdstaff Contributor Loading…
2 tasks done
Handle OpenAI tool_choice requests
#1070 opened Apr 25, 2026 by eloe • Draft
Support OpenAI stop sequences in server
#1069 opened Apr 25, 2026 by eloe Loading…
1
fix(gemma4): handle JSON-formatted keys in tool call parser
#1065 opened Apr 24, 2026 by afanty2021 Loading…
Add MiniCPM-V 4.6 support
#1058 opened Apr 24, 2026 by pzc163 Loading…
Add vision feature caching to all models
#1028 opened Apr 16, 2026 by Blaizzy Owner Loading…
6 tasks done
Expose presence_penalty, frequency_penalty, and per-penalty context_size on the server API
#1023 opened Apr 14, 2026 by esaruoho Loading…
1
refactor: improve model loading and resource handling in utils.py
#1019 opened Apr 13, 2026 by SyedaAnshrahGillani Loading…
fix: propagate the verbose to the Prefill tqdm
#1015 opened Apr 12, 2026 by PeterStaar-IBM Loading…
1
server: indicate finish reason properly when model made a tool call.
#1014 opened Apr 12, 2026 by viktike Contributor Loading…
perf: close 5.5% decode gap vs mlx_lm.server on streaming chat endpoint
#1012 opened Apr 11, 2026 by chilang Loading…
ProTip! What’s not been updated in a month: updated:<2026-04-04.