Agents should not edit blind.
Better lets them ask: “who worked on this file/symbol before, and is there a checkpoint I can reuse?”
Then parallel sessions compose into one release frontier.
Git import/export works today. Hosted sync coming soon.
https://t.co/ZG1aFdvS1E
@jpschroeder i think it will take time for people to catch up with this (and why i try to cater my content towards people a couple months behind), but definitely directionally agreed
Hydrogen brings major leaps in intelligence, speed, and reliability:
⚡ Cutting-edge intelligence: Surpasses Argon on the @ArtificialAnlys Intelligence Index at default high thinking (and climbs even higher at xhigh), while finishing tasks faster than competing models in its class.
💻 Cleaner coding & debugging: More than triples 3.8 Flash on Terminal-Bench 4.1, cuts fabricated claims by more than 5X, and adapts cleanly when a tool call fails or a search returns empty.
💼 Top marks on real-world work: Ranks #1 on Vals Tax Agent Bench, ties for #1 on Vals Legal Research, and leads on AutomationBench-AA, Multimodal Knowledge Agent (MKA), and Agents' Last Exam.
An open offer to Tesla: We fully support putting Cybercab first, and we’d be glad to take the Cybertruck line off your hands.
We’d fund it with a consortium of investors and run it at no cost to Tesla’s floor space, headcount, or focus. All of that goes to Cybercab. Cybercab gets first call on battery cells too. We’ll wait our turn.
Tesla gains incremental profit: a license fee, plus FSD and Robotaxi network revenues on every vehicle built.
We’d build the Cybertruck, a three-row CyberSUV, and an unmanned CyberCargo hauler. More autonomous miles, across every surface of this planet and beyond.
Worth a conversation?