Log inSign up
Image
user avatar
Prince Canuma
@Prince_Canuma
Apple MLX King 🤴🏽• Creator of (mlx-audio & mlx-vlm) • working on something new Ex-@arcee_ai • @neptune_ai • linktr.ee/prince.canuma
Krakow, Poland
github.com/Blaizzy
Born July 23
Joined July 2012
1,261
Following
22.5K
Followers
RepliesMedia
  • Pinned
    user avatar
    Prince Canuma
    @Prince_Canuma
    Jul 20
    Excited to introduce Nativ 🚀 Run frontier open models locally on your Mac. No accounts, no subscriptions, no cloud. ⚡ Built on mlx-vlm — multimodal + fastest on Apple Silicon 🔒 100% private — every token generated on your machine 🛠 Plug your coding agents into a local
    Image
    00:00
  • user avatar
    Prince Canuma
    @Prince_Canuma
    16h
    Hello world, from Nativ!
    user avatar
    Nativ
    @Nativ_AI
    17h
    Congrats to @liquidai on LFM2.5-2.6B! Day 0 support in Nativ 🎉 2.6B params · 128K context · built for agentic + coding workflows, fully on-device. On an M5 Max (48GB) with Nativ v0.2.2 — full bf16, no quantization: ⚡️ 11,231 tok/s prefill ⚡️ 82 tok/s decode 🧠 Full 128K
    Image
  • user avatar
    Prince Canuma
    @Prince_Canuma
    16h
    How fast is @liquidai LFM2.5-2.6B on Apple Silicon? We benchmarked it — from a single request up to 16 concurrent streams, all on one Mac. M5 Max (48GB) · Nativ v0.2.2 · full bf16 Get started 👉 blaizzy.github.io/nativ/
    Image
    Image
    user avatar
    Liquid AI
    @liquidai
    20h
    Today we release LFM2.5-2.6B, an agentic model that runs entirely on-device. It plans, calls tools, and works through multi-step tasks on phones, laptops, PCs, and robots. Data never leaves the device, and the marginal cost of each run is essentially zero. > Pre-trained on ~34T
  • user avatar
    Prince Canuma
    @Prince_Canuma
    16h
    Congrats to @liquidai on LFM2.5-2.6B! Excited to have partnered with them for Day 0 support in Nativ 🎉 Built for agentic + coding workflows — and it’s fast. On an M5 Max (48GB) with Nativ v0.2.2 — full bf16, no quantization: ⚡ 11,231 tok/s prefill ⚡ ~84 tok/s decode 🧠 Full
    Image