Skip to content
Source · Daily Brief

AI Morning Brief — 1 July 2026

Overnight, the US Department of Commerce lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5, clearing the way for their return on July 1 after weeks of restricted access. The news arrived alongside Anthropic’s release of Claude Sonnet 5 and the unveiling of Claude Science, making June 30 the company’s most active launch day in months.

Top stories

  • Fable 5 and Mythos 5 unblocked. The Department of Commerce notified Anthropic that export controls on its two most capable models have been lifted; Anthropic posted that access will begin restoring on July 1 and expressed gratitude to users for their patience. via The Verge, via TechCrunch
  • Claude Sonnet 5 released. Anthropic’s newest mid-tier model claims performance close to Opus 4.8 at lower prices, with stronger agentic capabilities and a new tokenizer that makes English text roughly 1.4× more token-dense. It shipped simultaneously to AWS Bedrock, GitHub Copilot, Cursor — where it scored 57% on CursorBench versus 49% for Sonnet 4.6 — and Perplexity. via TechCrunch
  • Claude Science launched. Unveiled at an event for pharmaceutical executives, the workbench bundles more than 60 preconfigured skills covering genomics and computational chemistry, a citation-verification agent, and local or HPC-cluster deployment so sensitive data never leaves the lab. Anthropic is positioning it as doing for scientific research what Claude Code does for software engineering. via MIT Technology Review
  • Google ships Nano Banana 2 Lite and Gemini Omni Flash. The image model generates results in under four seconds at $0.034 per image; Gemini Omni Flash brings text-to-video editing to the Gemini API for the first time at $0.10 per second. Google recommends chaining the two to take a concept from still to motion in a single session. via THE DECODER
  • Meituan releases LongCat-2.0. The 1.6 trillion-parameter MoE model — roughly 48 billion parameters active, trained on 35 trillion tokens — was built entirely on Chinese ASICs and supports a 1 million token context window; it is the model behind Owl Alpha on OpenRouter. The release is evidence that Chinese labs can reach near-frontier performance without Nvidia hardware. via THE DECODER

Who shipped

Anthropic dominated the cycle with three near-simultaneous moves: Sonnet 5, the Claude Science workbench, and the restored availability of Fable 5 and Mythos 5 following the export control lift. Google DeepMind added two generative media models in the same window, pushing into fast image generation and API-accessible video editing. Meituan posted the most significant open model release outside the US, demonstrating a credible path to frontier-class performance on domestic chips.

Open-source pulse

Meituan LongCat-2.0 is available now on OpenRouter. Huawei quietly open-sourced OpenPangu-2.0-Flash, a 92-billion-parameter MoE model with 6 billion active parameters and a 512K context window, with the larger Pro variant promised for July. DeepSeek DSpark, a new speculative-decoding framework that boosts per-user throughput by 60 to 85 percent on DeepSeek models, was released openly and has already been integrated into the SGLang inference engine.

Money, infra & hardware

Inference chip startup Etched came out of stealth with $800M raised and $1B+ in customer contracts at a $5B valuation, claiming SOTA throughput and power efficiency on inference workloads; first racks ship this summer. Autonomous-driving startup Wayve launched an $85M employee tender offer at an $8.5B valuation. Amazon announced a $1 billion forward-deployed engineering organisation to embed purpose-built agents within enterprise customers, following similar programmes at OpenAI and Anthropic.

Quiet corners

Mistral, Stability AI, and Cohere were absent from the primary-source feeds overnight. xAI posted no model-related announcements; the Grok account was quiet while the Elon Musk account remained focused on political commentary.

By the numbers

  • 420 stories in 24 h across 30+ sources
  • Most-mentioned model: Claude
  • Most-mentioned lab: Anthropic
  • Notable absences: Mistral, Stability AI, Cohere, xAI

Compiled by AI Feed’s editor from all 420 headlines collected in the 24 hours to 1 July 2026.