Retter Reboot #27

The Retter Reboot : Your weekly dose of ai & tech news

1. Anthropic's Fable 5 and Mythos 5 Are Back Online Worldwide

After a 19-day, government-ordered suspension, Anthropic's most capable Mythos-tier models returned to all users globally on July 1, once the US Department of Commerce lifted the export controls it had imposed on June 12. Access is restored across Claude.ai, the Claude Platform, Claude Code, and Claude Cowork, with cloud availability rolling back on. Anthropic is also proposing an industry-wide framework, co-developed with Amazon, Microsoft, and Google, for scoring how dangerous a given jailbreak is, to avoid future disproportionate shutdowns.

Source: Anthropic

2. Claude Sonnet 5 Launches as the New Default Model

Anthropic released Claude Sonnet 5 on June 30, its most agentic Sonnet yet, now the default for Free and Pro users and available across Max, Team, Enterprise, Claude Code, and the API. It narrows the gap with the flagship Opus 4.8 on reasoning, tool use, and coding while costing far less: introductory pricing of $2 per million input tokens and $10 per million output tokens runs through August 31. The positioning is deliberate, giving enterprises a cheaper way to run agents at scale as the company moves toward an IPO.

Source: Anthropic

3. California Signs a Statewide Deal With Anthropic

Governor Gavin Newsom signed a first-of-its-kind agreement giving every California state agency, city, and county access to Claude at a 50% discount, alongside a free training package. Departments from the DMV to healthcare services are already deploying Claude for customer service, case work, and cybersecurity. The move is notable given the simultaneous federal designation of Anthropic as a supply-chain risk, underlining how state and federal AI policy are diverging.

Source: Build Fast with AI

4. OpenAI Previews GPT-5.6 (Sol, Terra, Luna) Under Government Supervision

OpenAI began a limited preview of its new GPT-5.6 family, gated to roughly 20 trusted partners after coordination with the US government. Sol is the flagship for hard reasoning and cybersecurity, Terra the balanced mid-tier priced competitively with GPT-5.5 at half the cost, and Luna the fast, cheap option. OpenAI also plans to deploy Sol on Cerebras wafer-scale hardware in July, targeting up to 750 tokens per second, roughly 15x current serving speeds.

Source: OpenAI

5. Google Throttles Meta's Access to Gemini as Compute Runs Short

The Financial Times reported that Google capped Meta's use of its Gemini models because it could not supply the computing capacity Meta wanted to buy, a restriction in place since March that delayed several of Meta's internal AI projects. Google Cloud's undelivered order backlog nearly doubled to $460 billion in a single quarter, with Sundar Pichai admitting the company is compute-constrained. Meta has since accelerated its own model to reduce reliance on outside providers. The lesson: the best model means little without enough compute to run it.

Source: Cybernews

6. Anthropic Overtakes OpenAI in Business Subscriptions

Fresh market data shows Anthropic has passed OpenAI in self-reported revenue and business subscriptions, and is narrowing the gap in overall users. Anthropic said in May it was on course to hit $47 billion annualized revenue and reach profitability in 2029, a year ahead of OpenAI's projections. Meanwhile, ChatGPT's share of monthly generative-AI visits slipped below a majority for the first time, suggesting consumers are increasingly willing to switch between models.

Source: Fortune

7. South Korea Commits $880 Billion to AI, Chips, and Robotics

President Lee Jae-myung announced a national investment plan totaling 1,350 trillion won (about $880 billion) over 10 years, targeting semiconductors, AI infrastructure, and robotics. The bet reflects a growing conviction among nations that controlling AI infrastructure is the key to economic competitiveness in the coming decade, and adds another major state actor to the global compute race.

Source: Al Jazeera via Unrot

8. Five Eyes Warns Frontier AI Will Reshape Cyber Risk 'Within Months'

The Five Eyes intelligence alliance (the US, UK, Canada, Australia, and New Zealand) issued a rare joint statement warning that frontier AI models are set to fundamentally transform both offensive and defensive cyber capabilities. Their key line: the timeline is not years, it is months. This is a formal intelligence assessment that agencies are actively tracking capability thresholds in existing models, a signal for every organization to take AI-driven cyber risk seriously now.

Source: Build Fast with AI

Jul 3, 2026