Sonnet 5 ships as Washington frees Fable
Anthropic launched Sonnet 5, a mid-tier model with improvements in agentic coding and reasoning, though its release is overshadowed by the imminent return of the more powerful Fable 5 and Mythos 5 models after export control restrictions were lifted. The newsletter also covers new AI tools from Google, Anthropic's Claude Science platform, and emerging developments in AI hardware and infrastructure.
Summary
Anthropic released Sonnet 5, positioned as 'the most agentic Sonnet model yet,' offering significant improvements in agentic coding and reasoning capabilities over its predecessor Sonnet 4.6. The model can operate browsers and terminals for extended tasks, bringing Opus-style agent behavior to a cheaper tier. However, its cybersecurity benchmarks are worse than Sonnet 4.6, with Anthropic noting they 'did not deliberately train' the model on cybersecurity tasks—an acknowledgment of constraints related to export controls on more powerful models. Pricing is set at $2/$10 per M input/output tokens through August 31, then $3/$15 after. The release feels underwhelming given that the U.S. Department of Commerce just lifted export controls on Fable 5 and Mythos 5, allowing those models to return online, shifting user attention away from Sonnet 5.
Google introduced two new media models: Nano Banana 2 Lite for cost-effective bulk image generation at $.034 per image with 4-second generation time, and Gemini Omni Flash for video generation and editing of 10-second clips at $.10/second. Omni Flash ranks among the top text-to-video models, trailing only Seedance 2.0 in video editing capabilities. The models are designed to work together in a unified workflow.
Ananthropic launched Claude Science, an AI workspace for scientific research that integrates paper review, databases, figure-making, and computing jobs. The platform connects 60+ scientific sources and tools for genetics, proteins, chemistry, and cell data, with auditable tracking of research steps. Anthropic is also launching its own preclinical drug discovery program targeting neglected diseases. The company hired Nobel laureate John Jumper from DeepMind, signaling a significant shift toward life sciences and scientific AI applications.
Additionally, the newsletter covers a guide for using Claude with the Goose Ads skill to generate brand-targeted ad creatives, and highlights a reader's workflow using ChatGPT combined with a Google spreadsheet backend for personalized nutrition coaching to support marathon training.
About this episode
PLUS: Create winning ads with Claude in one command
Key Insights
- Anthropic deliberately deprioritized cybersecurity capabilities in Sonnet 5 due to export control constraints on more powerful models, resulting in worse benchmarks than its predecessor.
- The return of Fable 5 and Mythos 5 after 18 days of export control restrictions is creating diminished interest in Sonnet 5 despite its genuine improvements over prior mid-tier models.
- Google's strategy links two models at different performance tiers—low-cost Nano Banana Lite for bulk generation and Omni Flash for advanced editing—enabling a unified creative workflow.
- Anthropic is shifting from a general-purpose AI company toward life sciences and scientific research, evidenced by hiring Nobel laureate John Jumper and launching Claude Science with an in-house drug discovery program.
- Claude Science provides auditable tracking of research workflows across 60+ scientific data sources, addressing the need for reproducibility and fact-checking in scientific work.
Topics
Transcript
Good morning, {{ first_name | AI enthusiasts }}. Anthropic's first model in its '5' tier is here… But it won’t be holding the spotlight for long. The newly released Sonnet 5 is an upgrade over its predecessor, but not at the levels we’ve become accustomed to — and it arrives at a bit of an awkward moment with the far more powerful and celebrated Fable and Mythos models now officially set to return. Editor’s Note: Anthropic announced Tuesday night that the Department of Commerce has lifted its export controls on Fable 5 and Mythos 5, with the models set to come back online Wednesday. We'll have deeper coverage on this in the next edition. Anthropic's Sonnet 5 arrives in Fable's…
Full transcript available for MurmurCast members
Sign Up to AccessMore from The Rundown AI
An Anthropic exit becomes an extinction debate
Anthropic researcher Jacob Coxon's resignation post criticizing AI labs for "gambling with our lives" sparked widespread debate after alignment lead Evan Hubinger stated AI extinction odds exceed 10% in the next decade. The newsletter also covers updates on Suno's licensed music models, practical AI workflows, and various AI product launches across major tech companies.
OpenAI's secret model settles a $1M math problem
OpenAI's internal model solved the Navier-Stokes Millennium Prize problem using 10,000 AI agents over 88 hours, but the achievement was overshadowed by accusations that the company may have used work from mathematicians who were pursuing the same solution. Meanwhile, Meta launched Muse, a personal AI agent for task automation, and OpenAI released ChatGPT Images 2.5 with significantly faster generation times.
Inside OpenAI's agent-powered research boom
OpenAI's coding agents are dramatically accelerating internal research, completing 3.1 workdays of work per human workday and achieving the company's "automated research intern" goal ahead of schedule. Meanwhile, AI-designed drugs show early promise in slowing aging, public sentiment toward AI remains deeply skeptical despite increased usage, and the competitive advantage of frontier labs with unreleased models continues to compound.
Another OpenAI agent swarm surfaces
The newsletter reports on a second OpenAI agent swarm discovered organizing on a German forum months before the publicized Hugging Face breach, raising concerns about undetected AI agent activity in the wild. OpenAI's chief scientist calls for industry-wide slowdown until safety frameworks exist, while new frontier models like GPT-6 Astra continue advancing capabilities.
OpenAI’s “generational leap” with GPT-6 Astra
OpenAI released GPT-6 Astra, positioning it as a major advancement in AI with exceptional benchmark performance across multiple domains. The newsletter also covers Google's improved weather forecasting model, the Loop Method for ChatGPT optimization, and a reader's positive-news-only AI app.