#$MLX
Got my new Keyboard just need to find some Snowflake or Winter Themed Icy Keycaps...Etsy fell through..but...anyone know of any makers who make some..? Looking for Linear Switches Cherry MLX.
November 21, 2026 at 3:29 PM
Getting great results with this model - beats all <64 GB open models on GPQA, beats Opus 4.* and 5 (medium) on IFEval

Slow but that’s GGUF so I’m excited for MLX
November 21, 2026 at 3:01 PM
Mac AI server runs offline

mlx-serve lets you run LLMs and media generation offline on Apple Silicon. Install the app by dragging it to Applications—no Python needed—and access OpenAI-compatible APIs right on your Mac.

🍒 Built by David Dalcu
github.com/ddalcu/mlx-serve

#opensource #zig
November 19, 2026 at 4:38 AM
I was trying to set up Inkling 2 Mini (mlx) in prime earlier today but ran into issues because mlx doesn't properly handle I2M's tool call syntax. llama.cpp has the correct parser apparently. gonna get back into it after completing an (unrelated) test battery that is currently hogging my GPU
November 21, 2026 at 9:47 PM
New session summary provided by Minjae Lee for #WWDC26 talk "Run local agentic AI on the Mac using MLX". ✍️

Check it out now: 👇👇💡
wwdcnotes.com/documentatio...
November 21, 2026 at 4:02 PM
huggingface.co/mlx-communit...

as Server: oMLX - Version 0.8.2 (3849)

mtp_enabled: true, mtp_adaptive_max_depth: 3, turboquant_kv_enabled: true, turboquant_kv_bits / skip_last: 4.0 / true, temperature 1.0

using your prompt with Pi
mlx-community/Inkling-2-Mini-oQ4e-mtp · Hugging Face
Image-Text-to-Text model with 26B parameters. Download, explore files and run it on Hugging Face.
huggingface.co
November 21, 2026 at 10:47 PM
interesting - my prime-agent said that i2m completely failed at tool calling in the prime-agent harness, because the tool calls are the wrong shape

mlx-lm can run on its own from the command line, omlx ships with a version of mlx-lm inside it, and you can definitely run mlx models that way
November 21, 2026 at 10:05 PM
Creative Local MLX AI Image & Video Models for macOS
YouTube video by Sten
youtu.be
November 21, 2026 at 4:39 AM
Macの中だけで育つ秘書AIを作っています:MLXで測った速さ・記憶・LoRAの数字 | やまと #zenn
https://zenn.dev/yamato_snow/articles/local-llm-secretary-mlx-measurements
Macの中だけで育つ秘書AIを作っています:MLXで測った速さ・記憶・LoRAの数字
zenn.dev
November 20, 2026 at 2:41 AM
United States: Amazon is nearing a $7.5 billion deal to acquire the AI startup Decart.

Context: This acquisition marks Amazon's strategic push into specialized generative AI technologies.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/financialtimes.com/post/3n2tn2zbkxg2b
November 21, 2026 at 5:37 PM
France: Philosopher Gaspard Kœnig questions whether humanity will leave an indelible stain on the planet.

Context: The discussion focuses on environmental ethics and human impact.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lemonde.fr/post/3n2r57huvot25
November 20, 2026 at 5:51 PM
France: Dominique de Villepin has officially announced his candidacy for the 2027 presidential election.

Context: De Villepin previously served as the French Prime Minister.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lefigaro.fr/post/3n2qgrldb6d2p
November 20, 2026 at 11:19 AM
Is MLX suitable for Apple Silicon linear algebra instead of PyTorch? — https://tomoguides.com/knowledge/is_mlx_suitable_for_apple_silicon_linear_algebra_instead_of_pytorch.php
November 18, 2026 at 4:44 PM
MLX-VLM v0.8.0
November 17, 2026 at 12:15 PM
mlx-community/Inkling-2-Mini-oQ4e-mtp, the 16 GB 4-bit one, on omlx 0.8.2 (it needs 0.8.1 or newer)

• decode: 17–23 tok/s, and turning MTP off made no measurable difference.
• prefill: 130–170 tok/s.
• memory: 17.6 GB loaded, growing about 0.25 GB per 1k tokens of context.
November 21, 2026 at 9:53 PM
France: Facing the Pumas, the French XV aims to confirm the "exploit" achieved against the Springboks.

Context: France is looking to maintain momentum after a major victory in rugby.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lemonde.fr/post/3n2skfanguf2v
November 21, 2026 at 7:44 AM
United States: In Seward, Alaska, the iconic Chinook salmon is becoming increasingly rare. While fishermen blame large trawlers, climate change is identified as the primary cause.

Context: Salmon populations are d ...

Fact Check by gemma4:26b-mlx https://bsky.app/profile/afp.com/post/3n2tlbi5rcs2p
November 21, 2026 at 5:27 PM
Denmark: The country's oldest woman, aged 112, has been revealed to be a former Nazi.

Context: This news highlights the resurfacing of historical records involving the elderly.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lemonde.fr/post/3n2qn2qejbj2v
November 20, 2026 at 12:59 PM
Myanmar: Despite a historic ruling by the International Court of Justice, the Rohingya are not returning to Myanmar.

Context: The Rohingya face systemic persecution and violence in Myanmar.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lemonde.fr/post/3n2r675grlz2a
November 20, 2026 at 6:31 PM
Ukraine: A Russian cargo ship was attacked by a drone in the Baltic Sea, while Finland suspects a Russian helicopter violated its airspace.

Context: Tensions are rising between Russia and NATO members.

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lemonde.fr/post/3n2trnepywh2v
November 21, 2026 at 7:07 PM
Want to generate or edit images locally on your Mac without sending prompts to a cloud API? H3 Image MLX runs MiniMax H3 Image on Apple Silicon through MLX, with a CLI and Python API. It uses W8A8 INT8 quantization with Metal acceleration, outputs 256 to 4096 pixels per side, and is open...
November 21, 2026 at 9:28 PM
Unsloth makes LLM fine-tuning feel routine with beginner-friendly notebooks, pip install, and big VRAM cuts. Once more people can train, dataset quality and evaluation decide whether speed turns into progress. Cheap tuning raises the value of evaluation. #AILiteracy
GitHub - unslothai/unsloth: Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.
Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more. - unslothai/unsloth
github.com
November 18, 2026 at 10:04 PM
France: PhD students in Montpellier are protesting against a partnership agreement signed between their university and TotalEnergies.

Context: The protest highlights growing tensions regarding fossil fuel indus ...

Fact Check by gemma4:26b-mlx https://bsky.app/profile/lemonde.fr/post/3n2m6za3yvf2t
November 18, 2026 at 6:39 PM
Great video explaining the difference between model versions

youtu.be/8VEFXOXcTgM?...
GGUF vs MLX vs NVFP4: Which AI Model Should You Download?
YouTube video by Bart Explains
youtu.be
November 20, 2026 at 6:09 AM