Research, observations & things learned
Technology,
with the
notes attached.
A place to work through what we’re learning about AI and software. Technical explanations, thoughts from the work, and a closer look at the news.
Open the notebookSimple rules. Unexpected structure.
The reference desk
Choose the model.
Understand the machine.
Two guides to keep close when you’re exploring open models and running AI locally.
Capabilities · Licensing · Deployment
Open-source AI models
Compare model families, understand open-weight licensing, and find a starting point for evaluation.
Read the model guideMemory · Hardware · Local inference
Local LLM machines
Compare machine classes, memory limits, and the practical tradeoffs of running models on your own hardware.
Read the hardware guide01 / The long view
All technical guidesGet into
the details.
02 / In the notebook
What has our
attention.
New developments, questions worth asking, and our thoughts on where the technology is going.
All articlesNVIDIA Agrees to Buy Hugging Face. Open-Model Builders Should Read the Commitments.
NVIDIA's $12.93B Hugging Face agreement promises hardware choice. What the filing says, what remains unproven, and how builders can preserve portability.
GPT-6 Astra Brings Better Computer Use and a Higher Bar for Delegating Work
OpenAI's GPT-6 Astra improves computer use and complex work. A closer look at launch benchmarks, API costs, limited rollout, and safety monitoring.
Gemini 3.8 Flash Keeps the Token Price, but Task Costs Need a Fresh Look
Google launches Gemini 3.8 Flash and restricted Flash Cyber. More reasoning can raise task costs, and introductory API pricing ends in December.
Meta's Muse Spark 1.3 Takes Aim at the Cost of Supervising an Agent
Meta's Muse Spark 1.3 targets coding and long-task reliability. What its launch claims establish, and how to test whether it needs less supervision.
Fable 5.1 and Mythos 5.1 Share a Claude Model but Not an Access Policy
Anthropic's Fable 5.1 and Mythos 5.1 share a model but differ in safeguards and access. Cache pricing, enterprise privacy, and routing define the release.
Tencent's Hy4 Preview Is a 770B Apache Model With an API-Sized Deployment Problem
Tencent Hy4 opens 770B-parameter weights under Apache 2.0 and claims strong agent performance, but deployment remains data-center scale.
03 / Follow your curiosity
04 / Behind the notebook
People to
think with.
S5 Labs is where Joshua Wendt and Nick Flowers share learnings, thoughts, and current news. The writing is the starting point for a conversation.
More about usWe also work on AI systems, automation, and software.
How we can help