Skip to content

Pull requests: Blaizzy/mlx-vlm

New pull request New
38 Open 794 Closed
38 Open 794 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

perf(qwen3-omni): early-exit the thinker hidden-state extraction
#1462 opened Jun 28, 2026 by yuhuanowo Contributor Loading…
3
fix(qwen3-omni): eval mRoPE position_ids before the thinker forward
#1461 opened Jun 28, 2026 by yuhuanowo Contributor Loading…
2
fix(gemma4): recognize bare namespaced tool-call names
#1459 opened Jun 27, 2026 by SuperMarioYL Contributor Loading…
[codex] Skip read-only generation config defaults codex codex-automation
#1457 opened Jun 27, 2026 by Blaizzy Owner • Draft
Add Unlimited-OCR model support
#1427 opened Jun 23, 2026 by shuuul Loading…
9
Fix request-local MRoPE rope_deltas under continuous batching bug Something isn't working priority/p1
#1422 opened Jun 23, 2026 by scyyh11 Contributor Loading…
@Blaizzy
5
Add PP-OCRv6 model support
#1420 opened Jun 23, 2026 by mikolaj92 Loading…
2
Fix SAM 3.1 interactive point/box segmentation (prompt-insensitive noise)
#1409 opened Jun 19, 2026 by FAbrahamDev • Draft
[codex] Skip read-only generation config defaults codex codex-automation
#1399 opened Jun 19, 2026 by Blaizzy Owner • Draft
Fix diffusion detok marker preservation codex codex-automation
#1394 opened Jun 17, 2026 by Blaizzy Owner Loading…
2
Port glm_moe_dsa language model
#1390 opened Jun 16, 2026 by Blaizzy Owner • Draft
fix(nemotron_labs_diffusion): load MLX-format checkpoints via opt-in sanitize flag
#1386 opened Jun 16, 2026 by luckyayush777 Loading…
1
fix(gemma4): load legacy MLX checkpoints with obsolete shared-KV weights
#1382 opened Jun 14, 2026 by mvance Loading…
2
Fix token_context desync that crashes continuous batching
#1378 opened Jun 14, 2026 by rmaher001 Loading…
1
2
Minimax m3 support
#1374 opened Jun 13, 2026 by ivanfioravanti Contributor Loading…
9
Handle fully padded rows in Qwen 3.5 chunked prefill
#1373 opened Jun 12, 2026 by lucasnewman Collaborator Loading…
1
fix(server): honor per-request top_k/min_p (and seed) in the batched sampler
#1353 opened Jun 11, 2026 by mmmugh Loading…
fix(qwen3_omni): pass decoded audio arrays to the processor, not raw paths/BytesIO
#1352 opened Jun 11, 2026 by DoradusResearch Loading…
Add CoPE (Clipped RoPE) soft-clipping for zero-shot context extension
#1344 opened Jun 10, 2026 by machiabeli Loading…
4
Add MLX-VLM agent skills
#1343 opened Jun 10, 2026 by Blaizzy Owner Loading…
Add local client setup command
#1341 opened Jun 9, 2026 by Blaizzy Owner Loading…
Fix MTP server target model handling codex
#1319 opened Jun 7, 2026 by Blaizzy Owner Loading…
1
Port Nemotron H ModelOpt weights
#1314 opened Jun 6, 2026 by Blaizzy Owner Loading…
Add RealtimeSession: incremental audio for realtime interaction
#1289 opened Jun 4, 2026 by fum0passiv0 Loading…
feat(mla): FP8 KV cache for MLA models with absorbed-decode Metal kernel
#1284 opened Jun 4, 2026 by popfido Contributor Loading…
4
ProTip! Mix and match filters to narrow down what you’re looking for.