Yeah context rot is about bad tokens (mistakes etc) getting into the context and causing poor performance later on – what Theo is describing is more models straight up wasting tokens
INNOVATION
-
Building a Personal Product: From Problem to Solution
By
–
For the first time in a long time, I just built something *I* wanted. It's a really different feeling than building something other people want. I know the problem so well, and will be using the product from day one. This may become a new Otherside product… we'll see!
-

LMCache: LLM Engine Extension for Faster Response and Higher Throughput
By
–
LMCache an LLM serving engine extension to reduce TTFT and increase throughput, especially under long-context scenarios
-

JAM: Tiny Flow-based Song Generator with Fine-grained Control
By
–
JAM A Tiny Flow-based Song Generator with Fine-grained Controllability and Aesthetic Alignment
-
Dedicated Image Video Generating Models Show Strong Performance
By
–
Some of the deeicated image and video generating models have done pretty well, like this one: https://t.co/RTeWYSfdUl
— Simon Willison (@simonw) 29 juillet 2025Some of the deeicated image and video generating models have done pretty well, like this one:
-
GDB Talk and Latent Space Podcast Questions for Greg
By
–
also hit the bell to summon algo gods for the gdb talk pls thx! https://
youtube.com/watch?v=avWhre
BUYF0
… we're taking suggestions and questions for the @latentspacepod with him, if you have Open Model or GPT5 or anything AI Engineering + OpenAI that you specifically want to hear from greg, -

AI Index 2025 Report Reveals Shifts in Talent Investment Regulation
By
–
Want to understand AI's real impact? Last week, @StanfordHAI Executive Director @russellwald presented key insights from the #AIIndex2025 Report at @FortuneMagazine
's #BrainstormAI, revealing surprising shifts in talent, investment, and regulation: http://
bit.ly/4lktPYZ -
Building Custom AI Tools Without Built-in Tool Calling
By
–
This is what I do! I just build it myself instead of relying on built-in tool calling 🙂
-
GPT-5 reportedly being tested in ChatGPT platform
By
–
Possibly GPT-5 being tested in ChatGPT https://t.co/gb4UUQXjXh
— Peter Gostev (@petergostev) 28 juillet 2025Possibly GPT-5 being tested in ChatGPT
-
Optimized Prompts for Advanced AI Models
By
–
Oh, I meant that that looked like an optimized-for-o3 prompt!