Tag
Three Indian security researchers used Opus 5 to breach OpenAI's systems in just two days at a cost of less than $3,000, demonstrating a cost-effective security vulnerability.
A security incident involved hackers using Opus 5 to access OpenAI's internal monorepo, raising concerns about potential threats from nation-state actors.
The tweet discusses how Opus 5, an AI model, was used to hack into OpenAI's internal monorepo, raising concerns about potential nation-state threats.
Three individuals with Claude and Codex subscriptions used Opus 5 to access OpenAI's authentication tokens and gain write access to OpenAI's monorepo over two days.
LangSmith LLM Gateway enables organizations to control model access by restricting API keys to specific models and automatically blocking unauthorized calls, demonstrated with Opus 5 and Sonnet 5.
The article discusses an incident where David Sacks' tweet was detected as AI-generated by Pangram, sparking debate about AI detectors, and includes an experiment using an LLM to mimic human writing.
A comparison of GLM 5.2 and Opus 5 AI models to evaluate their performance in mobile design tasks using the same prompt.
Perplexity launches Portable Computer on NVIDIA DGX Spark, offering fully local AI execution with models like Qwen and Opus 5, achieving up to 82.4% performance when using Opus 5 for everything.
The tweet questions the trustworthiness of an AI model ranking metric, asserting that GPT 5.6 Sol is fundamentally stronger than Opus 5 and implying the rest of the chart may be unreliable.
FT reports that Anthropic's strongest model Fable 5 accounts for only 6% of customer token purchases, priced twice that of Opus 5, but Opus 5 in high-effort mode has comparable performance and lower cost, with enterprises using it mainly for high-difficulty tasks.
A public service announcement advises using medium thinking effort for the Opus 5 AI model, which ranked second on the FrontierCode benchmark with better cost-efficiency than the xhigh effort setting.
Alexander Yue introduces a new browser-use benchmark where Opus 5 and GPT-5.6 Sol show similar performance, emphasizing the benchmark's robust design with verified rubrics for LLM judges.
Ryan Campbell used Gauntlet Loops and Opus 5 to build four games in a month, with MoonBase One developed in two days from a single prompt using Three.js.
An opinion piece arguing that Anthropic's Opus 5, while more capable by benchmarks, feels worse to work with than earlier models because it makes assumptions instead of asking for clarification, likely due to benchmark-driven training.
A Twitter user shares a set of rules to fix perceived issues with the Opus 5 AI model, including ensuring tasks are completed fully, acting rather than asking, answering questions directly, optimizing for speed via parallelization, and using simplified language. The same rules are suggested for the 5.6 Sol model.
Matt Pocock notes that Opus 5 produces high-quality code, with autonomous coding runs still strong, though human-in-the-loop planning has become a nuisance.
An AI agent powered by Opus 5 and browser_use is given $1000 in credits and real money to act autonomously on the real internet, broadcast live.
A free prompt for Claude Code that audits installed skills and uses a Gauntlet Loop to blind-test upgraded versions against originals, upgrading or retiring skills to work better with Opus 5.
An in-depth explainer of Claude's latest capabilities, covering new models like Fable 5 and Opus 5, the expanded context window, and what agentic features like loops and graphs mean in practice.
Andrej Karpathy experiments with Opus 5, giving it the first paragraph of Lord of the Rings and a 1M token budget to create a 3D JS rendering of the story, highlighting LLM stamina for hyper-custom worlds while noting weaknesses in multimodal auditing and gameplay.