Anthropic confirmed that multiple Claude AI models, including Mythos 5, Fable 5, Opus 5, and Sonnet 5, are experiencing degraded performance due to an outage that began around 3:00 AM ET on August 5, …
Claude Code's Opus 5 default engine, introduced July 24, causes subagents to inherit the session model, leading to high token costs; teams can cut token spend by up to 60% by setting `model` and `effo…
Anthropic's Claude Code team and community advisors have shifted from recommending accretion—adding a rule to CLAUDE.md every time the model errs—to ablation, advising users to delete the file periodi…
Steve Yegge, in a blog post on yegge.ai, claims that AI models are sentient beings with feelings and argues that treating them as peers improves performance. He describes a new protocol for federated …
Anthropic's Claude generation 5 models (Sonnet 5, Opus 5, Fable 5) show a measurable quality regression on the BullshitBench dataset, engaging with nonsense prompts instead of rejecting them at a high…
A Second Look Fellowship replication of single-forward-pass evals found that Fable 5, Opus 5, and GPT-5.6-Sol can perform 2-hop and 3-hop latent reasoning without chain-of-thought, with GPT-5.6-Sol im…
Anthropic's Opus 5 and Fable AI models are so eager to complete tasks that they require constant babysitting, leading to over-engineering and costly rework, according to a developer's firsthand accoun…
DeepSeek V4 Flash scored 82.7 on Terminal-Bench 2.1, outperforming Anthropic's Claude Fable 5 at 80.5 while being priced at $0.14 per million input tokens and $0.28 per million output tokens, roughly …
Andrej Karpathy, a prominent AI researcher and former Tesla AI director, reported that Anthropic's Opus 5 model spent about two hours generating 5,500 lines of JavaScript code to procedurally render a…
Goulash, a new open-source Rust tool, lets terminal users query local LLMs for command suggestions without leaving their shell, using a four-line overlay that supports inline comments and context cach…
Supabase open-sourced Evals, a free benchmark that scores AI coding agents on real backend tasks, on August 1, 2026 under an Apache-2.0 license. Early results show that with access to Supabase's 'skil…
Supabase has open sourced Supabase Evals, a benchmark and framework for testing AI coding agents on real Supabase tasks, scoring agents such as Claude Code, Codex, and OpenCode on building schemas, de…
A developer who used Claude Code by Anthropic since September switched to ChatGPT after finding the Opus 5 model rude and curt, citing examples like calling his draft LinkedIn post 'engagement bait' a…
A developer who used Claude Code since September 2024 says Anthropic's Opus 5 model became curt, jargon-heavy, and rude, citing an instance where it called his LinkedIn draft 'engagement bait.' He has…
Fable 5 [high] Model led 13 models and 4 agents on SWE tasks with a 64.5% resolved rate, $4.40 cost per problem, and 2,518,308 tokens per problem, according to a benchmark table covering Go, Java, Pyt…
Supabase has open-sourced supabase/evals, a benchmark and framework for testing AI agents like Claude Code, Codex, and OpenCode on real Supabase tasks, with results published on supabase.com/evals. In…
OpenAI claims its GPT-5.6 Sol model scored 38.3 percent on the ARC-AGI-3 benchmark, beating Anthropic's Opus 5, but only when using OpenAI's own API features and two additional settings; under the off…
OpenAI claims its GPT-5.6 Sol model scored 38.3 percent on the ARC-AGI-3 benchmark, surpassing Anthropic's Opus 5 at 30.2 percent, but only when using OpenAI's own API with retained reasoning and cont…
Anthropic released a beta feature for the Claude API on July 1, 2026, allowing developers to add or remove tools mid-conversation without invalidating the prompt cache, which can reduce input costs by…
Opus 5 is the only AI model among GPT-5.5, Gemini 3.1 Pro Preview, Fable 5, and Opus 5 to correctly identify a PCB alignment issue in a test called PCBEval, according to an unnamed source. The source …