Because of an Azure mistake, some API resellers have already opened up access to GPT-6.2 Sol and GPT-6.2 Astra. This is absolutely insane...
(I checked, and it's real, but it seems to have been accidental!)
The launch has never been this close!!!!
Gemini 4 Hydrogen by @GoogleDeepMind is the fastest model in the Agent Arena top 10.
A typical Hydrogen step takes 7.3 seconds, about 2.7x faster than Claude Sonnet 5.5 and 3.1x faster than GPT-6.1 Sol.
Faster steps mean less time waiting on tasks. Across Agent Arena sessions, Hydrogen spends under 10 minutes on 82% of its tasks, compared with 55% for GPT-6.1 Sol and 44% for Claude Sonnet 5.5. Within 30 minutes, that rises to 97%, versus 81% for Sol and 78% for Sonnet 5.5.
And it isn't a quality trade-off: Hydrogen's net improvement is on par with both models.
How we measure it: A step is one model API call in an Agent Mode task. All times run from the first model request to the last streamed reply, without tool execution.
You can now connect multiple local computers to the same dot and choose which one it uses for a task.
It can build and test your app on a dedicated machine while you keep coding on your laptop.
Meta AIが小型効率モデル「Muse Ember」を発表。複雑な推論実行やアプリ横断ワークフロー、ツール連続呼び出し、コード修正などに特化し、1Mトークンコンテキストとマルチモーダル対応を備える。
Haiku 5.5と同水準の性能を、約1/3のタスク単価で実現し、Lunaを上回る効率を出す小型モデル
Today we're introducing Muse Ember, our smallest and most efficient model, built for complex reasoning execution and everyday agent work.
Muse Ember can carry workflows across apps, call tools in sequence, run tests and make targeted code fixes. With a 1M-token context window and native understanding of text, images, audio and video, it can also work across large document sets and varied inputs.