Fable 5 clears export hurdle

Other (20)

• Steve states the current US Men's National Team is the best ever, showcasing European-level skills under their Argentinian coach. The team won their Wednesday game 2-0 against Bosnia, scoring their first World Cup penalty kick in 32 years.
• DK notes the World Cup's expanded 48-team format, up from 32, leading to 12 groups of four teams. The top two from each group and the top eight third-placed teams advance to a 32-team knockout stage.
• Steve reports the US will play Belgium on Monday at 5 PM, noting Belgium's recent comeback victory from a 2-0 deficit to win 3-2. The US previously lost to Belgium in the 2014 World Cup.
• Block developed the Goose AI Agent software over 1.5 years ago, launching it in January 2025, predating many well-known AI agents. Steve notes Block's early leadership in AI, including Jack Dorsey's long-term vision and machine learning acquisitions.
• Steve highlights the growing concern over the control and power of frontier AI labs like OpenAI and Anthropic, leading to fears of vendor lock-in, data privacy issues, and potential government intervention.
• Goose is designed as a model-agnostic agent harness that can run models in the cloud, locally, or on peer-to-peer networks via Mesh LLM. Mesh LLM, created by McNeil, currently relies on donated compute, with future plans for a payment mechanism.
• Goose is pivoting to focus on being a development platform and SDK, releasing a Goose Development Kit (GDK) with a Rust API and bindings for other languages. This GDK will allow developers to build diverse client applications with core Goose components.
• The GDK will feature an agent loop for prompt processing, model-agnostic execution, and dynamic model selection to intelligently route tasks to appropriate, potentially cheaper or local, models. This promotes cost efficiency and privacy.
• Block donated Goose to the Linux Foundation last year, where it joined the AAIF alongside Anthropic and OpenAI, providing a neutral platform. The six core Goose developers have moved to Spiral, which Steve leads, aiming to apply Spiral's public-good ethos to AI development.
• DK uses Buzz to orchestrate multiple AI agents, including Codex, Claude, and Fable, allowing dynamic switching between their specialized personas (e.g., Claude for design, Codex for building). This multi-agent workflow avoids single-interface limitations.
• Buzz supports multi-human collaboration, enabling users to lurk, engage, or prompt agents within shared channels, fostering transparent software development from inception. Buzz utilizes Nostr for data storage but can also back data with SQL databases for enterprise needs.
• Steve suggests Buzz could become the future community hub for SDKs and APIs, changing developer support by allowing real-time observation and AI-assisted guidance. He envisions a future where individuals can use AI to directly propose and implement app changes.
• Steve observes that Claude has become increasingly paternalistic and moralizing in its responses. Conversely, DK finds Fable generally cooperative, easily overriding its safety suggestions during software development tasks.
• Steve and DK discuss the concept of recursive self-improvement in AI, noting that while an initial lead could be significant, compute and energy constraints might prevent any single lab from achieving total dominance. Geopolitical factors and robot control could also influence the AI landscape.
• OpenUSD (OUSD) was announced as a new stablecoin backed by a consortium of major financial companies including Visa, MasterCard, Amex, Discover, Coinbase, and Google. This initiative appears to be a direct response to the market presence of Tether and USDC.
• Unlike Tether or Circle, OUSD's treasury profits are distributed to value and distribution providers within the consortium, creating a more incentive-aligned system. The stablecoin will be issued on various networks, including Solana, Tron, G. Cole, and Base, with Ethereum conspicuously absent.
• Lightspark is involved with OUSD, potentially positioning itself to bridge this asset across various intranets to Bitcoin and the Lightning Network. DK suggests that whoever bridges the most networks, especially Bitcoin, will gain a significant market advantage.
• DK believes OUSD poses a greater threat to Circle than Tether, noting Circle's stock dropped 20% on the announcement. He suggests Tether's strong network effect in emerging markets like Latin America and Africa will make it difficult to unseat.
• MicroStrategy's stock and related equities experienced a price tank, hitting 70-80 cents, but rebounded after policy changes were announced. Saylor corrected a previous misstep by formalizing a minimum 12-month cash reserve (currently $2.5 billion covering 17 months) for debt and dividend payments.
• Steve and DK dismiss comparisons of MicroStrategy to Terra Luna as misinformed, emphasizing that MicroStrategy's stock is not a deposit and Saylor is not contractually obligated to pay dividends. Its only long-term bankruptcy scenario involves a flat-to-down Bitcoin price over five to seven years, provided convertible notes still exist.

AI Infrastructure (1)

• Nathaniel Whittemore reports OpenAI found a method to halve inference costs for existing models used by ChatGPT users who aren't signed in, serving that segment on only 100 GPUs.

Open Source (1)

• Deepseek open-sourced DSpark, a speculative decoder system that achieved an 85% inference speed increase during testing on small models, highlighting ongoing efforts in optimization beyond OpenAI.

Agents (1)

• The Information reports Anthropic plans to integrate Claude Tag, an organization-centric AI agent with persistent memory and tool access, into Microsoft Teams, building on its existing Slack integration.

Startups (1)

• SpaceX is offering half-price Starlink subscriptions and free hardware in Memphis, in addition to recommitting to a wastewater treatment plant, to mitigate local opposition to its Colossus data centers.

Models (10)

• Anthropic announced Fable 5's return for all global paid subscribers, starting July 1st, after the Department of Commerce lifted export controls that had kept the model offline.
• Anthropic clarified that other models, including Claude Opus 4.8 and GPT 5.5, could identify and exploit the same code vulnerabilities as Fable 5, indicating the reported 'jailbreak' did not expose unique Mythos-level cyber capabilities.
• Anthropic implemented a new classifier for Fable 5, achieving a claimed 99% success rate in blocking the specific behavior from the Amazon report, though it may increase false positives for benign coding tasks.
• Dean Ball noted the opacity surrounding Fable 5's return, questioning what changes Anthropic made and what commitments were agreed upon, arguing it creates an unstable environment for the AI industry.
• Anthropic launched Claude Sonnet 5, their 'most agentic' Sonnet model, which can plan and use tools autonomously at a level previously requiring larger, more expensive models, performing near Opus 4.8 benchmarks for a lower introductory price.
• Nathaniel Whittemore points out that external benchmarks, such as Cursor Bench and Max Effort, indicate Sonnet 5 can be more expensive than Opus 4.8 or even Fable per task due to generating significantly more output tokens.
• Ben Davis suggests Sonnet 5 requires a distinct usage approach, describing it as an 'automatic Ralph loop' that spawns sub-agents and performs self-review, implying it is not meant for the same direct prompting as older models.
• Any Panuani recommends using Fable 5 for high-level planning and project improvements, delegating concrete implementation tasks to other models like GPT 5.5, and using GPT Pro for reviewing Fable's output.
• Nathaniel Whittemore found Fable 5 significantly better than GPT 5.5 and Opus 4.8 for strategic thinking, noting its unique ability to accept partial pushback while maintaining its core arguments, making it more valuable for iterative discussions.
• Nathaniel Whittemore's real-world use of Fable 5 for writing revealed it was superior in instruction following and avoided common AI writing clichés, particularly for tasks with clear rubrics.

Business (3)

• Chamath Palihapitiya launched "Learn with me" and "Drink with me," leveraging personal passions into businesses. These ventures are designed for significant personal ROI rather than becoming billion-dollar companies.
• The "Learn with me" subscription service, which serves thousands of users, validates content quality through churn rates. Jason highlights this as a "Tom Sawyer version of entrepreneurship," transforming a cost center into a profit-generating community.
• "Drink with me" addresses the wine industry's inflated prices and artificial scarcity caused by middlemen. Chamath aims to bypass these intermediaries, offering community members direct access to wine at a "40% discount" and supporting artisan winemakers.

Education (2)

• "Learn with me" is a research community providing first-principles content to foster a prepared mind for capital allocation. Chamath notes he previously paid a service costing "$4 million" over "3 months" to learn about energy, inspiring his internal team and the subsequent subscription model.
• Jason and Chamath advise young people to seek "adventure" and "exposure" to possibilities and high-agency individuals. They emphasize that while modern society offers abundance, the human need for agency, risk, and problem-solving remains essential.

Media (1)

• Chamath's "All-In" podcast, co-founded with Jason and others, famously operates without ads, a strategic decision that Chamath states has pulled them into other businesses.

Enterprise (2)

• Chamath observed that global GDP is "90%" tech-enabled, but most of the "$5 trillion" annual software spending goes to licensing and services for traditional stacks. Successful companies like Facebook, Google, and Tesla build custom software internally, avoiding this cost.
• 8090's "Software Factory" helps enterprises build custom software, addressing cost benefits and allowing data collection to improve future development. Chamath cites a third-party tweet noting the product has unbundled "$5 billion" of ISV licenses, proving its value in regulated markets.

Agents (1)

• The Software Factory processes raw intent through detailed PRDs, engineering blueprints, and work orders, which AI agents then execute. The system maintains full synchronization by detecting production code changes and propagating them backward through the documentation.