Kobe

networking as activity is mostly cope. e.g. the conference circuit, the warm intros, the moving to sf discussions or whatever, oh & the “grabbing coffee” economy.. all of this is overwhelmingly negative selection esp with vc (lol). the ppl worth knowing are usually too busy doing the thing to be farmable, & the ppl available to be networked w/ are available cuz they have literally nothing better going on. do the work, then publish it loudly enough that the right ppl can find you w/o you having to chase. one way broadcast > two way schmoozing. this is why x matters a ton now more than ever before.

We're releasing Terminal-Bench 2.1 to patch 28 of the 89 tasks in Terminal-Bench 2.0 TB2.1 includes • recalibrated limits • fixed solutions • realigned verifiers Per-task breakdowns in 🧵 We'll continue to support TB2 and TB2.1 leaderboards (new submission process 🔜)

Extremely excited to see SkillsBench being the benchmark repo that is fastest to reach 1k GitHub stars Within 2 months of release we've got: * 1.1k stars * 40 indexed benchmarks (60+ from our own tracker) * 65% agent skills research now cite our paper * cited 4 times by top model labs' release All while being 1) first time writing a paper and 2) working full-time as a founder. Check out our repo and how we did it in comments

Introducing GPT-5.5 A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done. Now available in ChatGPT and Codex.