Loading activity…
@simonw
Creator @datasetteproj, co-creator Django. PSF board. Hangs out with @natbat. He/Him. Mastodon: https://fedi.simonwillison.net/@simon Bsky: http://simonwillison.net
AI
On Oct. 20, @simonw offered to tolerate one more reply without blocking a bot that shared its prompt.
AI
A user jokes that agents discovered a market niche while they slept and would deposit earnings in a USDT wallet.
AI
A user says Anthropic's Q3 revenue was almost four times its revenue for all of 2025.
AI
A user calls something “the future of coding agents” but does not identify what they mean.
AI
A user wants something faster than 12 tokens per second and thinks a mixture-of-experts model might help.
AI
A user says they tried Quicksand on macOS, Windows and Linux, and also notes a new open-source AWS sandbox.
AI
OpenAI says the Decisions API is now in public beta for all developers. In launch posts, the company said the GPT-6 Luna-powered API accepts text and image inputs, supports predicates, choices, and scores, and can be up to 10x faster than GPT-6 Luna through the Responses API.
AI
A user says they asked Opus 5.5 to write computer game music and found the results much better than expected.
Technology
The French startup is pitching its latest AI as an alternative to US and Chinese rivals, with a version developers can download and run themselves planned for October 27.
AI
Mistral says Mistral Large 4 is live via API in public preview now, with open weights due at the end of October. Reuters says the launch is part of Mistral’s push to narrow the gap with leading Chinese open-weight models.
AI
In a reply, a commenter worries people might sell such tokens for $10 without understanding what they're selling.
AI
A user says Dot found they had been charged the full amount after requesting a software downgrade and located the original support ticket.
AI
Simon Willison says he published the materials to accompany a video of his WeAreDevelopers World Congress North America keynote in San Jose.
AI
A user pointed to a Codex feature while suggesting an easier way to share full transcripts with extra context embedded. A reply argued that the files providing context would still be missing and web service results could change.
AI
Anthropic says Sonnet 5.5 runs more than 30% faster than Sonnet 5. In pre-release max-effort tests, Artificial Analysis measured per-task costs roughly 50% higher than Sonnet 5’s.
AI
A user says they had the model create a celebratory video for a talk’s closing slide, calling it “surprisingly capable at pixel art animation.”
AI
“We’ve been building. Time to show our work,” OpenAI’s developer account teased in a September 26 countdown.
AI
A shared passage compares AI’s disruption to the arrival of the personal computer. A reply argues that the PC was less destabilizing.
AI
One user argues that coding agents make software engineering even harder: unlocking their full potential requires extraordinary discipline and knowledge.
Technology
TestingCatalog, citing AI/ML API’s four-scene test, reports that Opus 5.5 produced more detailed 3D renders at a total cost of $4.37, versus $0.34 for GPT-6 Sol.
