# Capability Articoli collegati

Il Centro Notizie HTX fornisce gli articoli più recenti e le analisi più approfondite su "Capability", coprendo tendenze di mercato, aggiornamenti sui progetti, sviluppi tecnologici e politiche normative nel settore crypto.

GPT-5.6 Suffers 'Brain Surgery', Fable 5's Return to Be a Weaker Version?

GPT-5.6 and Fable 5 Face "Lobotomization" and Restrictions OpenAI has launched its new GPT-5.6 model in three limited tiers: Sol (advanced), Terra (balanced), and Luna (fast/economical). However, access to the most powerful Sol version is severely restricted, reportedly due to U.S. government intervention over safety concerns linked to its capabilities. Similarly, Anthropic's Fable 5 was globally disabled for 72 hours after a demonstration revealed its potential to exploit security vulnerabilities. It is now set to return in a heavily restricted "neutered" version for select partners only. Industry experts warn that such stringent safety "guardrails" could lead to "lobotomized" models with diminished intelligence and creativity. Developers are concerned about restricted access, potential extra costs, and models that no longer match their original promised capabilities. This marks a shift towards controlled access, contrasting with the earlier era of widely available, powerful AI.

marsbit06/29 08:37

GPT-5.6 Suffers 'Brain Surgery', Fable 5's Return to Be a Weaker Version?

marsbit06/29 08:37

Anthropic Released the "Most Powerful Model," But Most People Can't Use It

In April, Anthropic launched a preview of its "Mythos" model, which was not publicly released due to its exceptional ability to autonomously discover high-risk zero-day vulnerabilities, posing a security threat if misused. It was restricted to a trusted group of security partners under "Project Glasswing." On June 10, Anthropic officially released Fable 5 and Mythos 5. They share the same underlying model but are distributed under different rules. Fable 5 is for general users, while Mythos 5 remains locked for trusted security partners. Benchmarks show Fable 5 leading in software engineering and long-task execution, with significant improvements in generating production-ready code. However, Fable 5 includes a safety classifier that automatically downgrades requests related to cybersecurity, biochemistry, or model distillation to the weaker Opus 4.8 model. This mechanism, while intended for safety, can affect the user experience and has faced criticism for being overly conservative. Pricing is another key point. Fable 5's API costs are double that of Opus 4.8. Furthermore, after a free trial period ending June 23, it will be removed from standard subscription plans, requiring users to purchase additional credits for access. This shift signals a move towards pay-as-you-go pricing for the most advanced capabilities. The strategy highlights a growing divergence in the AI industry: while some players like DeepSeek are drastically cutting prices, Anthropic is increasing them for its top-tier model, using cost as a filter for high-value users. The article suggests the AI market is stratifying, with commoditized capabilities becoming cheaper while premium, cutting-edge models command a significant price premium.

marsbit06/10 23:52

Anthropic Released the "Most Powerful Model," But Most People Can't Use It

marsbit06/10 23:52

What Are Some Good Paths for Chinese Web3 Entrepreneurship? (Part 5)

This article explores pathways for Chinese Web3 teams to pivot toward AI, building on a previous discussion. It focuses on two specific team profiles: **Security & Risk Control Teams:** These teams, skilled in smart contract auditing, wallet security, and on-chain monitoring, can transition to providing **Agent behavior auditing and AI security governance**. As AI Agents automate tasks, access data, and trigger payments, enterprises will need solutions to monitor permissions, audit logs, control data access, and prevent anomalies—creating a strong B2B demand. **Application & Community-Focused Teams:** Instead of completely rebranding as AI companies, these teams should use AI to **enhance their existing products**. For example, research platforms can use AI to summarize information and identify signals; community tools can automate user support and analysis; and educational products can create personalized learning paths. The key is integrating AI to solve existing user pain points, like information overload or high operational costs. The article also advises against certain AI directions for Chinese Web3 teams, such as building general-purpose large language models (too resource-intensive), creating overly broad Agent platforms (hard to monetize), developing AI traders/automated yield products (high regulatory and risk sensitivity), or simply adding superficial AI features without genuine value. The core conclusion: Successful migration depends not on chasing AI hype, but on **identifying how a team's existing Web3 capabilities—be it in data, payments, security, or user operations—can address real needs in new AI application scenarios.**

marsbit06/04 14:53

What Are Some Good Paths for Chinese Web3 Entrepreneurship? (Part 5)

marsbit06/04 14:53

Anthropic's 'Cry' Sparks Wall Street Panic! 27-Year-Old, Mythos Instantly Defeated by 8 AIs

Anthropic's announcement of Claude Mythos, an AI tool claiming to discover thousands of zero-day vulnerabilities—including a 27-year-old bug in OpenBSD—triggered alarm on Wall Street and prompted emergency meetings among financial regulators fearing systemic cyberattacks. However, independent tests revealed significant exaggerations in these claims. Researchers found that many reported vulnerabilities were in obsolete software or were impractical to exploit, and the findings relied on only 198 manual reviews. Furthermore, multiple smaller, open-source AI models (some with as few as 3 billion parameters) successfully identified the same critical flaws at a fraction of the cost, demonstrating that AI cybersecurity capability does not linearly scale with model size. Meanwhile, users reported severe performance degradation in Claude Opus 4.6, with reduced reasoning depth and increased API costs. Critics, including prominent hacker George Hotz, accused Anthropic of overstating risks for marketing purposes, creating a "wolf cry" scenario where hype overshadows reality.

marsbit04/12 08:03

Anthropic's 'Cry' Sparks Wall Street Panic! 27-Year-Old, Mythos Instantly Defeated by 8 AIs