Tech • AI • Robotics
Fable 5.1, OpenClaw 2, ChatGPT and Solaris: a crazy return!
9.4/10TL;DR
Fable 5.1, ChatGPT Ads, fresh European scrutiny of OpenAI, Open Clow 2, and Runway Solaris all point to the same shift: AI is moving from novelty tools to infrastructure for work, commerce, regulation and interfaces.
KEY POINTS
Fable 5.1 targets enterprise work
Anthropic has released Fable 5.1 as a higher-performing model aimed at long, structured professional tasks rather than casual chat. It is positioned for code refactoring, migration, debugging, large-scale financial analysis and knowledge-heavy support systems. The model can work for hours on asynchronous tasks, supports a 1 million-token context window and up to 128,000 tokens of output, enabling large reports and complex multi-step jobs.
Performance rises while pricing shifts on cache
Standard pricing remains at $10 per million input tokens and $50 per million output tokens, keeping the model among the most expensive commercial offerings. The major change is cached input pricing, which is now four times cheaper than on Fable 5. That makes repeated enterprise workloads, retrieval-augmented generation and chatbot deployments over large proprietary knowledge bases far more economical.
A “digital employee” positioning emerges
The practical appeal of Fable 5.1 lies in reliability and persistence more than in benchmark wins alone. It is designed to revisit sources, maintain coherence across long documents and reduce back-and-forth corrections, though that rigor can also mean slower responses and higher token use for even small edits. The model is visible in Claude interfaces but consumes additional paid credits rather than standard lower-tier subscription usage.
Security claims are central to the launch
Anthropic is presenting Fable 5.1 as powerful enough to detect software vulnerabilities without being willing to exploit them. A more restricted companion version is reserved for selected partners, reflecting growing industry attempts to separate high-capability assistance from offensive misuse. The test for the market will be whether those guardrails hold under real-world pressure.
ChatGPT advertising has reached France
OpenAI has expanded ChatGPT Ads into France and broader Europe, bringing in-conversation advertising to free and low-cost users. Paid subscribers are excluded, while free and low-tier usage is being monetized through display and click-based ad formats. The system mirrors the long-standing web search model: AI access for users, advertising revenue for the platform.
OpenAI opens a self-serve ad platform
Advertisers can now buy campaigns directly through ads.openai.com, with a reported minimum budget of $15 per day. Campaigns can be billed by impressions or clicks, product catalogs can be imported for automated ad generation, and early access conditions suggest low competition may currently make placement unusually cheap. The platform remains in beta, with signs of unstable metrics and incomplete reporting.
Ad revenue is already material
OpenAI has reportedly generated $1 billion in ad revenue in roughly 200 days in the United States, against a stated target of $2.5 billion. That early performance suggests conversational AI advertising could quickly become a normalized business model. The key commercial question is whether user trust can be preserved once monetization sits inside the flow of dialogue.
Europe places ChatGPT under DSA-style oversight
The European Commission has designated ChatGPT as a very large online search engine, alongside Reddit and Roblox, triggering obligations under the Digital Services Act. The move is significant because ChatGPT is not primarily presented as a search engine, yet its search-like usage appears large enough to cross the 45 million monthly users threshold that activates the regime.
Audits, transparency and potential fines now loom
OpenAI now faces recurring internal audits, outside scrutiny and mandatory reporting on systemic risks, including illegal content, effects on minors, mental well-being, fundamental rights and electoral processes. The company has four months to comply, with potential fines of up to 6% of global turnover for failures. Enforcement will largely depend on national regulators, notably in Ireland, where many major tech firms are based.
Open Clow 2 pushes shared agent workflows
Open Clow 2, released under the MIT license, expands agentic work by allowing persistent cloud sessions that multiple people can share. Instead of a single user running an AI helper on one local machine, teams can collaborate on the same session, assign parallel tasks and merge outputs, including code changes and repository work. That marks a step toward genuinely multi-user AI coworkers.
Runway Solaris hints at post-code interfaces
Runway has unveiled Solaris, an experimental interface system that generates interaction flows in real time rather than relying on traditionally coded UI behavior. The system interprets mouse actions and visual context to infer what a user wants next, closer to a dynamic visual assistant than a fixed application. Its limits remain major, especially with text handling and accessibility, but the concept could challenge how websites, apps and software interfaces are built.
CONCLUSION
The latest wave of releases shows AI moving deeper into enterprise operations, advertising markets, regulatory frameworks and software creation itself. The common issue is no longer whether the tools are impressive, but how quickly institutions, businesses and developers can adapt to what they make possible.
Explain this