X / Twitter
Sam Altman, OpenAI CEO
Altman self-deprecatingly assessed GPT-5.5 against competitors: its frontend interfaces may lag (“gets looksmaxxed”), but its reasoning intelligence dominates (“IQmog hard”), and he promised to address the frontend issue. He also reposted amazement at how a competing product could code so quickly, showing interest in code-generation speed competition. The post received 2574 likes.
https://x.com/sama/status/2048160404376105179
https://x.com/sama/status/2048165186482389253
Ryo Lu, Cursor Designer
Lu announced he had fully switched to GPT-5.5 plus Composer 2, calling it a perfect combination of intelligence, speed, and cost. He stressed that the configuration is exclusive to Cursor and currently available only there. The post received 629 likes, indicating resonance among developers.
https://x.com/ryolu_/status/2047879353313431649
Peter Yang, Roblox Product Manager
Yang demonstrated GPT-5.5 with Codex by building a Star Fox game after 15 minutes of prompting. He observed the AI spontaneously testing its own game, an interesting instance of AI self-playtesting. He also admitted frequently asking Codex to repair his OpenClaw configuration, humorously portraying everyday AI-toolchain use.
https://x.com/petergyang/status/2047882568520102226
https://x.com/petergyang/status/2047880605204144159
https://x.com/petergyang/status/2048136116474437869
Peter Steinberger, OpenClaw Founder
Steinberger praised GPT-5.5’s role-playing “character game,” calling it a clear capability leap. He also cheerfully said OpenClaw was overwhelming GitHub’s servers and thanked the GitHub team for cooperating, showing its scale creating infrastructure challenges.
https://x.com/steipete/status/2047871519762567468
https://x.com/steipete/status/2048185940267380815
Nan Yu, Linear Product Lead
Yu offered counterintuitive adoption advice: users at the technical frontier suffer most because tools and practices change rapidly, continually disrupting workflows. The best position is “one or two steps behind,” adopting at scale after tools and paradigms stabilize so proven methods can solve real business problems.
https://x.com/thenanyu/status/2048172709075402787
Zara Zhang
Zhang described writing’s essence: sitting down to type is the last step, not the beginning. Real writing happens mentally and constitutes 80% of the process. Writing therefore is not equivalent to generating words, and even powerful AI text generation cannot replace that thinking. This further explains why she does not write with AI.
https://x.com/zarazhangrui/status/2048101948826751429
Dan Shipper, Every CEO
Shipper shared an epistemological insight: we never observe a model itself, only its behavior under our way of asking. This matters for evaluation because different questions can produce radically different conclusions about capabilities.
https://x.com/danshipper/status/2048026505461080076
Swyx
Swyx previewed an ongoing project with @hhua_, to be released after ICML in Seoul, suggesting followers reserve calendar time. He also shared a discussion of why “code is not cheap” and an interesting comparison of PM and engineer thinking on the same problem.
https://x.com/swyx/status/2048126046487322835
https://x.com/swyx/status/2047948378659754373
Nikunj Kothari, FPV Ventures Partner
Kothari shared recently followed X accounts spanning texture, harness, aura, design, AI, and other areas, encouraging others to follow them. He also observed that concision and clarity are what most writing lacks. Good writing distills complex ideas into a form readers need to think about, rather than dumping vague thoughts on them.