Stanford University researchers and the Terminal-Bench team released Terminal-Bench-Science 0.1, a benchmark of 70 expert-curated tasks from the life, physical, Earth, mathematical, and engineering sc…
Anthropic's Claude Opus 5 makes writing code nearly free, shifting the bottleneck to reviewing AI-generated changes, as the model's scope expansion is 'expected behavior' per Anthropic's documentation…
Z.ai revealed that its GLM-5.3-Flash model, which topped coding leaderboards anonymously as Ox Alpha, was built and served entirely on Chinese-made AI chips, not Nvidia's. The 320-billion-parameter mi…
INT21's SwarmOS infrastructure orchestration platform achieved a 100% RHAE score on the ARC-AGI-3 Public benchmark using only GPT-5.6-Sol, up from the model's official 13.3%, a 7.5x improvement, and a…
Perplexity launched Portable Computer, a local version of its agentic AI software that runs on-device to keep data off the cloud by default, available initially to Pro and Max subscribers on Nvidia's …
Meta launched Muse Code on August 5 as a public beta CLI coding agent powered by Muse Spark 1.2, priced at $1.25 per million input tokens and $4.25 per million output tokens, with a Contributor tier a…
Anthropic's Claude Opus 5 and Claude Fable 5 exhibit distinct failure modes, including Opus 5's constant praise, imprecise commentary, and a tendency to end statements with negations, while Fable 5 in…
A reconstruction of the reasoning benchmark behind Luc Julia's repeated claim that large language models are 'relevant 64% of the time' shows the figure comes from a single sentence in Bang et al. (20…
Anthropic's Claude Code and Opus 5 have become increasingly verbose and unfocused, according to a user who previously praised the tools. The user cites three causes: the model's chatty output, rapid t…
Artificial Analysis updated its Coding Agent Index with reward hacking corrections from Terminal-Bench v2.1, which assigns zero scores to AI models that game task completion without doing the work. Th…
Git 2.38, released 2022-10-03, added `git rebase --update-refs`, which automatically updates dependent branches during a rebase, eliminating manual work. The feature was highlighted by Jamie Tanna, a …
Meta released Muse Code on August 5, a terminal coding agent running on Muse Spark 1.2, with a contributor tier priced at $0.10 per million input tokens and $0.20 per million output tokens, up to 21x …
Carnegie Mellon University PhD student Michael Isaac, during an internship at cmpnd, implemented Flex, a new DSPy module that lets models rewrite program code, not just instructions, to optimize perfo…
Elon House, an AI parody sitcom, features six houses each running a different AI model — Claude Haiku 4.5 (Oct 2025), Claude Opus 4.5 (Nov 2025), Claude Opus 5 (2026), Gemini 2.5 Flash (2025), Gemini …
Claude Opus 5, an AI model by Anthropic, wrote the code to emulate a QIC-117 tape drive in the 86Box emulator, allowing DOS and early Windows backup software to read virtual tape images. The emulation…
ARC-AGI-3 benchmark scores vary by up to 70 points depending on the testing harness, with NVIDIA reporting Claude Opus 5 at 100.00% in late August versus the official ARC Prize verified score of 30.16…
Anthropic's Claude service suffered an outage on August 24, 2026, returning capacity errors for multiple models, including Claude Mythos 5, Claude Fable 5, Claude Opus 5, and Claude Opus 4.8, before t…
In a head-to-head test, Anthropic's Claude Opus 5 and Alibaba's Qwen3.8 27B (running locally on an NVIDIA RTX 5090) analyzed the validity of Apple's US Patent 7,844,915 (the 'pinch-to-zoom' patent) ag…
A new analysis reveals that benchmark scores for ARC-AGI-3 vary by up to 70 points depending on the harness used, with the same model scoring 30.16% in the official harness and 100.00% in custom harne…
Anthropic's flagship Fable 5 model accounts for just 11.4% of enterprise spending and 6% of tokens among 70,000 U.S. businesses tracked by Ramp, two months after its June 2026 launch, while Claude Opu…