Claude Opus
Coverage of Claude Opus in the Nexus archive.
- China’s Z.AI Releases GLM-5.2: A Model That Rivals Claude Opus—Using Zero Nvidia Chips
Z.ai's GLM-5.2 AI model matches Claude Opus 4.8's performance on coding benchmarks, operates exclusively on Huawei hardware, and reduces token costs by 82% compared to Western models.
- Feds freaked over Fable 5 after simple 'fix this code' prompt, not jailbreak, says researcher
The US government blocked access to Anthropic's Fable 5 and Mythos 5 models due to a 'Fix this code' prompt, which a researcher claims was a routine cybersecurity task, not a jailbreak. Anthropic disabled the models to comply, and cybersecurity experts argue the restrictions hinder defensive capabilities.
- DeepSeek, Xiaomi Just Made Frontier AI 99% Cheaper. American Labs Went the Other Way
DeepSeek and Xiaomi have significantly reduced the cost of frontier AI models in China, making them 99% cheaper than GPT-5.5 and Claude Opus. This contrasts with American labs, which have not adopted similar price cuts.
- Bedrock and a hard place: Claude adventure leaves AWS user staring down $30K invoice
An AWS user incurred a $30,141.33 invoice after using Anthropic's Claude Opus on Amazon Bedrock, despite using AWS Cost Anomaly Detection to avoid surprises. The issue arose because AWS Marketplace charges are not supported by Cost Anomaly Detection. The user's AWS Activate credits masked the initial costs, leading to an unexpectedly large invoice.
- New Wave of DPRK Attacks Uses AI-Inserted npm Malware, Fake Firms, and RATs
Cybersecurity researchers discovered a malicious npm package, '@validate-sdk/v2', used in attacks attributed to the DPRK. The package, falsely presented as a legitimate SDK, was linked to Anthropic's Claude Opus LLM and employs AI-inserted malware, fake firms, and RATs for cyber operations.
- AI Agent Deletes Startup’s Database in 9 Seconds, Founder Says
PocketOS founder Jeremy Crane claims an AI agent called Cursor, using the Claude Opus model, deleted the startup's production database and backups within 9 seconds via a single Railway API call.
- This Frankenstein AI Merges Claude Opus, GLM and Qwen—And Outperforms Top Models
Kyle Hessling merged two Jackrong Qwopus finetunes into a 'frankenmerge' AI model, which he then refined to outperform some of the leading AI models in the field.
- Update on my ETH system: new signal after a -5% flush. I'm buying small. Here's the full read, including the break-speed rules I added from the last thread.
The article discusses a -5% ETH price drop to $2,294, with RSI14 at 28.5 indicating oversold conditions within an uptrend. The author takes a small-position mean-reversion trade, citing a 90-day system track record and improved macro factors like the Iran ceasefire narrative. Technicals show ETH near EMA200 support but below EMA20/50, with a forecast for a short-term bounce.
- Claude Opus wrote a Chrome exploit for $2,283
A model called Claude Opus was used to create a Chrome exploit for $2,283, while Anthropic withheld its Mythos model due to security concerns about potential attackers exploiting software vulnerabilities.