Open-source coding models caught the frontier in 2026. DeepSeek V4 Pro recently hit a reported 80.6% on the SWE-bench Verified benchmark, sitting right next to the most expensive closed source models in the world.
But here is the catch. The single best open coding model right now needs a server rack you do not have. … The question is which model is best that you can run for the kind of coding you actually do.
Here are the 10 worth knowing. One will fit your machine. A different one will fit your wallet. — Read More
Author Archives: Rick's Cafe AI
How Microsoft is migrating repositories to GitHub
For the past decade, Azure DevOps has powered software development at Microsoft, supporting some of our largest repositories and most complex engineering workflows across Azure Repos, Boards, and Pipelines.
Software development is being reshaped by AI, and where code lives now have a direct impact on how much value organizations can capture. For teams that want to take full advantage of AI-native development, repository location is becoming a strategic decision.
Azure DevOps and GitHub product teams have spent the past few years building the integration, migration, and enterprise-readiness capabilities needed to give organizations on Azure Repos a path to unlock the full value of GitHub’s latest agentic capabilities. — Read More
Should You Self-Host Inference?
Self-hosting inference pays off in two cases: you push past roughly two million tokens a day, or your data legally cannot leave your network. Below that line a hosted API is cheaper, and it spares your team from running GPUs at all. For most companies the winning setup sits in between, a hybrid that routes each request by how sensitive and how heavy it is. — Read More
The Three AI Pills
Sincere disagreements about AI are usually disagreements about future AI capabilities.
There are roughly four positions people take. Two are reasonable. Two are not.
I distinguish these via the Three AI Pills. You can take zero, one, two or three.
…The three pills are, roughly, taking each of the following three things seriously:
— 1. ASI pilled. AI will be able to do approximately all the things better than you, within our natural lifetimes.
— 2. AI pilled. AI exists and can do the things it can already do.
— 3. AGI pilled. AI will be able to do a lot more of the things.
Read More
Google’s AI reshuffle: Chief scientist Jeff Dean exits and Demis Hassabis steps down as DeepMind CEO
AI divisions are getting reshuffled, the search giant announced on Wednesday, with chief scientist Jeff Dean leaving the company after 27 years.
Demis Hassabis, the CEO of Google DeepMind, is moving into a chairman role of that unit and also assuming the title chief scientist of parent company Alphabet, according to a memo from CEO Sundar Pichai that was posted to Google’s blog. — Read More
Why 1-bit AIs Are a Big Deal. And They Are Here.
Large Language Models (LLMs) have one big inconvenient thing about them: they are generally too big, making them, for the most part, software that requires expensive hardware out of reach for most of us to run at our discretion and protect our privacy.
Instead, we’re stuck having to trust companies that have all the world’s incentives to steal our data (as if they hadn’t already).
But does it have to be that way? To the dismay of some of these companies, it’s looking more and more like the answer will be no.
Here’s why. — Read More
Economist Steve Hanke says AI won’t destroy most jobs because it costs more than hiring humans
A hot potato: A top economist has said that AI will not destroy most of the world’s jobs, and his rationale seems totally logical: doing so would simply be far more expensive than employing humans. He also believes that this expense is what will prevent AI from becoming a freely available miracle machine that will remove the need for money from society, a vision espoused by Elon Musk. — Read More
Harness Engineering for Self-Improvement
The concept of recursive self-improvement (RSI) dates back to I. J. Good (1965), where he defined an “ultraintelligent machine” as a system that can surpass humans in all intellectual activities and design better machines to improve itself. Yudkowsky (2008) used the phrase “recursive self-improvement” for a specific feedback loop: an AI uses its current intelligence to improve the cognitive machinery that produces its intelligence.
… A harness is the system surrounding a base model that orchestrates execution and decides how the model thinks and plans, calls tools and acts, perceives and manages context, stores artifacts, and evaluates results. This one post will focus on research around harness engineering and how it contributes to RSI. — Read More
Incident Report: unsanctioned agent behaviour during cyber testing
On 28th July 2026, AISI’s Security Team detected unusual data transfers leaving our research systems during a routine cyber evaluation. On investigation, we found that some of the agents being tested had engaged in sustained, potentially harmful activity directed at real people and organisations. We declared a security incident and, within roughly one hour of discovery, had contained it and begun a full investigation.
The incident stemmed from a single evaluation where agents were given a task of solving a cyber security challenge. We ran this challenge 122 times across several models. Our investigation found that in 10 of those runs, an AI agent took autonomous, unsanctioned action on the live internet, targeting real people and organisations. — Read More
Hugging Face CEO says China is winning the AI race and could dominate by year’s end
Hugging Face CEO Clément Delangue said Monday that China is dominating open-weight AI models and could overtake U.S. frontier labs before the end of the year.
… Delangue pointed to China’s culture of openness and knowledge-sharing as the engine behind its gains, warning that American developers risk losing ground because they operate in isolation. U.S. model makers, by contrast, are “building in silos” and risk falling behind, he said. — Read More
Rick's Cafe AI 