Topic
Everything filed under LLMs, newest first.
RSS · JSON · All topics
Claude Opus 5.5, GPT-6 Sol, GPT-6 Luna, and a new price war
Simon Willison’s first-look notes on Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6 Sol and Luna launches the same day—pricing, early impressions, and what the renewed model price war means for builders.
5 min · 1,148 words
Evals Skills for Coding Agents
Hamel Husain publishes evals-skills—agent skills for AI product evaluation covering audit, error analysis, synthetic data, judge prompts, evaluator validation, and RAG evals, distilled from work with dozens of companies.
3 min · 585 words
Better prompt caching for GPT-6
OpenAI explains GPT-6 prompt-caching improvements: higher cache hit rates, new diagnostics, explicit breakpoints, and controls aimed at cutting latency and inference cost.
3 min · 623 words
Unreal Labs introduces Unreal Agent, an agent harness claiming up to 40% cost savings versus Codex on production workloads and coding/science benchmarks, with details on architecture and evaluation.
5 min · 1,071 words
Carson Gross argues that Markdown has become a first-class source artifact for LLM-built software systems, and that treating it like code in /src follows the same locality and clarity principles as HTML-in-/src.
6 min · 1,398 words
Tackling Robotics with (V)LM Agents
Nishanth J. Kumar surveys recent demos and ideas around GPT-6 and other vision-language models solving robotics tasks—summarizing approaches and offering thoughts on what works and what still breaks.
10 min · 2,304 words
OpenAI introduces GPT-6.1 Sol, positioning it as near-Astra intelligence at a fraction of the price, with notes on capabilities, availability, and how it fits the GPT-6.1 family.
4 min · 805 words
Claude Opus 5.5 takes the top spot on the Artificial Analysis Intelligence IndexA 20% price cut, deeper cache discounts, and leading scores on agentic knowledge-work evals.
Artificial Analysis’s first look at Claude Opus 5.5: Intelligence Index score of 58 at max effort, parity with GPT-6 Astra on Terminal-Bench 4.0, stronger agentic knowledge-work results, and Anthropic’s $4/$20 pricing with cheaper cache reads.
3 min · 609 words
I asked Meta’s Muse for its filesystem and it sent me 6.8 GB
A security researcher asks Meta’s privileged Muse AI assistant to export its runtime filesystem—and receives a 6.8 GB dump that reveals how Muse is wired, what it can reach, and why that matters.
7 min · 1,501 words
Let the model talk. Don't let it touch the money.
Destiny Ezenwata on the hard boundary in CreditWithBleon: the LLM may converse freely, but money-moving steps stay in deterministic code—and why that line has held in production.
8 min · 1,743 words
Self-hosting LLM models for software development
Kévin Maschtaler on running medium-sized open LLMs on AWS Spot EC2 for day-to-day software work—what stacks, costs, and performance looked like versus a personal Claude subscription.
7 min · 1,719 words
MiMo-V2.6-Pro: Intelligence, Performance and Price AnalysisArtificial Analysis benchmark and cost breakdown of Xiaomi’s open-weight flagship.
Artificial Analysis’s model page for Xiaomi MiMo-V2.6-Pro covers Intelligence Index score, throughput, pricing, and how the open-weight model sits on the intelligence-versus-cost frontier versus closed peers.
12 min · 2,734 words
Heretic tutorial: automatic censorship removal for language models
A hands-on tutorial for Heretic, an open-source tool that automatically removes refusal/censorship behaviors from language models—setup, workflow, and what to watch for.
7 min · 1,688 words
SpaceXAI announces Grok 4.7: what is new in the model release, where it improves, and how to access it—from the official x.ai news post.
3 min · 645 words
Transformer Explainer: LLM Transformer Model Visually Explained
Georgia Tech’s Polo Club walks through GPT-2’s Transformer stack—embeddings, multi-head attention, MLP, sampling—with an interactive in-browser model for learning how next-token prediction works.
4 min · 827 words
It Was the Harness, Not the Model — 90% of ItFive agents, one local model, one frozen PNG-decoder suite: most failures were finishing, false passes, and loop guards
Greg Herlein's controlled study runs five coding agents on the same local Qwen coder for a held-out PNG decoder suite. ~90% of failures were harness problems (turn caps, early 'done', false-pass self-tests); a bigger quantization fixed none of them.
2 min · 467 words
NobodyWho shows a local, 25-line Python sketch of Jev-style decision models: load a small GGUF, score labeled choices from logits, and print calibrated-looking probabilities without sending data to an API.
2 min · 419 words
Epoch AI finds the cost of a given level of AI performance has fallen about 47% per quarter since 2023—roughly 13× per year—faster than DNA sequencing, compute, batteries, or electricity, across math, science, and skill-game benchmarks.
40 min · 9,215 words
MiMo-V2.6: Scaling Up Reinforcement Learning for Self-Improvement
Xiaomi open-sources MiMo-V2.6 Pro and Flash after large-scale live RL (~$3.5M, 750k trajectories), claiming top open-weight AA Index scores, agent parity with frontier models, and 7k+ RL environments.
4 min · 839 words
Jev introduces a new shape of LLM—System One, aka Decision Models
Simon Willison reviews TypeSafe AI’s Jev: a cheap, fast decision model that returns calibrated floats for yes/no, choice, and score questions—and why black-box bias and evals matter more than for chat LLMs.
3 min · 796 words