Gemma 4 12B および 31B は、Mac 上の完全にオンデバイスの Apple Core AI で実行されるようになりました 🎉 Apple 独自のオンデバイス ランタイム、組み込みモデルと同じ API で実行されます。ファーストパーティ モデルと同様に、それらをアプリにドロップできます。 12B ~42 tok/s・31B ~16 tok/s (M4 Max)
本文
Gemma 4 12B & 31B now run on Apple Core AI, fully on-device on Mac 🎉 They run in Apple's own on-device runtime, same API as the built-in models. You can drop them into your app like a first-party model. 12B ~42 tok/s · 31B ~16 tok/s (M4 Max) http://github.com/john-rocky/coreai-model-zoo
Gemma 4 12B & 31B now run on Apple Core AI, fully on-device on Mac 🎉 They run in Apple's own on-device runtime, same API as the built-in models. You can drop them into your app like a first-party model. 12B ~42 tok/s · 31B ~16 tok/s (M4 Max) https://t.co/fTjdSwXgjv https://t.co/…
white pill for my nerds: 60fps e-ink display a random guy outperformed entire eng teams by developing a pixel by pixel driver for e-ink displays that makes it 60fps. he did that after work for months, launched it yesterday. the future is bright
white pill for my nerds: 60fps e-ink display a random guy outperformed entire eng teams by developing a pixel by pixel driver for e-ink displays that makes it 60fps. he did that after work for months, launched it yesterday. the future is bright https://t.co/FERYA8iIjA
AMD CEO LISA SU は、235B モデルを実行し、月額 440 ドルの AI スタックを置き換えるミニ PC をステージ上で披露しました amd の ryzen ai max+ 395 は、1 枚のシリコン上で 2,000 億のパラメータ モデルを実行する初の x86 チップです。 CPU と GPU は 128 GB のユニファイド メモリを共有し、個別のグラフィックスは不要
本文
AMD CEO LISA SU HELD A MINI PC ON STAGE THAT RUNS A 235B MODEL AND REPLACES YOUR $440/MONTH AI STACK amd's ryzen ai max+ 395 is the first x86 chip that runs a 200 billion parameter model on one piece of silicon. cpu and gpu share 128gb of unified memory, no separate graphics https://twitter.com/starmexxx/status/2065859150140072261
AMD CEO LISA SU HELD A MINI PC ON STAGE THAT RUNS A 235B MODEL AND REPLACES YOUR $440/MONTH AI STACK amd's ryzen ai max+ 395 is the first x86 chip that runs a 200 billion parameter model on one piece of silicon. cpu and gpu share 128gb of unified memory, no separate graphics htt…
Apple の新しい Core AI (WWDC'26) を、iPhone 17 Pro 上のオンデバイス LLM の MLX および CoreML に対してベンチマークしました。その結果は私を驚かせました。 Qwen3-0.6B デコード速度 (tok/s): • Core AI (GPU、パイプライン): 180 Warm 🥇 • MLX (GPU): 115 • Core AI (ANE): 50 • CoreML-LLM (ANE): 39
本文
I benchmarked Apple's brand-new Core AI (WWDC'26) against MLX and CoreML for on-device LLMs on an iPhone 17 Pro. The results surprised me. Qwen3-0.6B Decode speed (tok/s): • Core AI (GPU, pipelined): 180 warm 🥇 • MLX (GPU): 115 • Core AI (ANE): 50 • CoreML-LLM (ANE): 39
I benchmarked Apple's brand-new Core AI (WWDC'26) against MLX and CoreML for on-device LLMs on an iPhone 17 Pro. The results surprised me. Qwen3-0.6B Decode speed (tok/s): • Core AI (GPU, pipelined): 180 warm 🥇 • MLX (GPU): 115 • Core AI (ANE): 50 • CoreML-LLM (ANE): 39 https://…