K2-Think a reasoning system that achieves frontier performance with just a 32B parameter model, surpassing or matching much larger models such as GPT-OSS 120B and DeepSeek v3.1 vibe coded a chat app for it in anycoder
GENERATIVE AI
-
Claude API Code Execution Tool Features Compared
By
–
That one isn't a feature of the http://
Claude.ai consumer-facing app, it's only available to users of their API https://
docs.anthropic.com/en/docs/agents
-and-tools/tool-use/code-execution-tool
… I expect the new user-facing feature uses the same underlying architecture though -
Transparency in LLM Usage: Sharing Prompts and Transcripts
By
–
I generally don't use custom rules because I don't like to get an experience that's different from the "default" everyone else gets – it makes it harder to write about things. When I write about Codex or GPT-5 or other LLMs I always share my exact prompts and transcripts
-
AI Overview Livestream: GPT-5, Models, Video, Agents, Robotics
By
–
Directo terminado. Casi 3 horas dando un repaso general de la actualidad de la IA (GPT-5, modelos frontera, imágenes, vídeo, agentes, robótica y mucho más). Y una visión compartida de hacia dónde se mueve la Inteligencia Artificial Lo tenéis ya resubido en Youtube!
-
Service Quality Disparity Between $10 and $10,000 Monthly Plans
By
–
Why do you expect service for paying $10/mo Service you get if you pay $10,000/mo
-
Claude Code Interpreter Successfully Recreates Charts from Screenshots
By
–
I re-ran the chart recreation from a screenshot and XLSX file experiment in the brand new Claude Code Interpreter feature (against Claude Sonnet 4) and got satisfactory results from that as well https://
simonwillison.net/2025/Sep/9/cla
ude-code-interpreter/#something-much-harder-recreating-the-ai-adoption-chart
… -
Anthropic’s New Feature Name Gets Heavily Criticized
By
–
Here's proof that Anthropic really did pick a legendarily bad name for this new feature
-
AI Labs Struggle with Feature Naming and Explanation
By
–
I'll stop making fun of AI labs for sucking at naming things when they stop sucking at naming things https://
simonwillison.net/2025/Sep/9/cla
ude-code-interpreter/#ai-labs-find-explaining-this-feature-incredibly-difficult
… -
ChatGPT Feature Comparison with Anthropic’s Implementation
By
–
Yeah, it's been one of my favorite features of ChatGPT for over two years now, I'm glad to see Anthropic copy it more closely. Here are my previous notes on it:
-
Claude generates Python chart code from uploaded XLSX file
By
–
It's very good! It took a little bit of prompting but I managed to get it to recreate the chart on the left using Python and matplotlib against an XLSX file that I uploaded to Claude