The ARC Prize Foundation reported that Anthropic's Opus 5 scored 30.2% on the ARC-AGI-3 benchmark, measuring action efficiency against a median human player, not problems solved. The independent bench…
Perplexity is not shutting down and its revenue is climbing, with annualized recurring revenue topping $450 million in March 2026, up 50% in a single month, and more than 100 million monthly active us…
Anthropic's per-model prompting pages serve as a maintenance list for instruction files like CLAUDE.md, which become stale after each model release because new models interpret prompts more literally,…
A staged Windows reproduction of the ChatGPT desktop app's Codex view measured a window with two images opening in 21.7 seconds and a window with 40+ images opening in 313.7 seconds, according to an i…
ChatGPT's desktop app, Codex CLI, and IDE extension can cause local CPU, RAM, and disk spikes during coding tasks. A July 24, 2026 guide from ChatGPT documentation recommends seven free fixes, startin…
An AI terminal agent can find available domain names by writing and running a script that checks each candidate against the live Verisign registry, returning only unregistered names. There are 163.6 m…
Six adversarial critic subagents, dispatched in parallel by a Claude Code blog-post skill, returned two skill_adjustments that named the instruction file section responsible for a flaw, enabling a sel…
Anthropic's Claude Fable 5, released June 9, 2026, leads the SWE-Bench Verified leaderboard with a 95% issue resolution rate, far ahead of Claude Opus 4.8 at 88.6%. The model's 1M-token context window…
SkillsBench, an arXiv benchmark published in February 2026, found that curated skills boosted average task pass rates from 33.9% to 50.5% across 87 tasks and 18 model-harness configurations, with heal…