ANTHROPIC LAUNCHES CLAUDE SCIENCE, AIMS TO COMPRESS DRUG DISCOVERY TIMELINES 10X
AI REPORT WIRE SERVICES | Filed 7:45 AM EST | Source: Pharmaceutical Technology
Anthropic has launched Claude Science, a dedicated application built for scientific research workflows, with an initial focus on drug discovery, protein structure analysis, genomics, and computational biology.
"Compress life sciences R&D cycles by a factor of 10."
The launch is part of a stated goal from Anthropic CEO Dario Amodei to dramatically shorten the time between a research question and an actionable result in the life sciences. Unlike a general chat assistant retrofitted for lab work, Claude Science is purpose-built around scientific data formats and workflows from the ground up — the clearest signal yet that frontier labs see vertical, task-specific apps as the next front after general assistants.
↑ back to top
HERMES AGENT BECOMES GITHUB'S FASTEST-RISING AI PROJECT THIS WEEK
Open Source Desk | Source: GitHub (NousResearch/hermes-agent)
Hermes Agent, built for long-term, personalized AI collaboration rather than one-off Q&A, is this week's fastest-rising AI repository on GitHub. It learns a user's habits and adapts to their workflow over time, positioning itself as an ongoing collaborator rather than a disposable chat session — a design bet that stands in contrast to most agent frameworks, which reset context between tasks.
↑ back to top
OPENKNOWLEDGE, AN OPEN-SOURCE AI-FIRST NOTION ALTERNATIVE, HITS SHOW HN
Apps Desk | Source: Show HN
OpenKnowledge debuted on Show HN as an open-source, AI-first alternative to Obsidian and Notion, built around writing and sharing plain markdown files with built-in integration for Claude and Codex. It's a small but telling entry in a growing category of Show HN launches this cycle: tools that skip building a new chat surface entirely and instead wire AI directly into formats people already use.
↑ back to top
OLLAMA SPEEDS GEMMA 4 NEARLY 90% ON APPLE SILICON
Local Models Desk | Source: Ollama Blog
Ollama shipped a performance update on July 1 delivering nearly 90% faster token generation for Gemma 4 on Apple Silicon, powered by multi-token prediction and automatic tuning, alongside tightened model loading and refreshed MLX and llama.cpp engines. The release also added smarter launch/runtime detection, including thinking-capability detection and auto-install support for Claude Code and opencode.
↑ back to top
ANTHROPIC OVERTAKES OPENAI IN SELF-REPORTED REVENUE
Business Desk | Source: Fortune
Anthropic says it has overtaken OpenAI on self-reported revenue and expects to be profitable in 2029, a year ahead of its rival's own projections. The company said in May it was on track to hit $47 billion in annualized revenue. The claim lands alongside mounting industry criticism of token-based pricing: Palantir CEO Alex Karp has publicly blasted the model both companies rely on, arguing enterprises are souring on "tokenmaxxing" as AI costs surge.
↑ back to top
ANTHROPIC MODELS BACK ONLINE AFTER TWO-WEEK GOVERNMENT STANDOFF
Policy Desk | Source: Bloomberg
Anthropic's Fable 5 and Mythos 5 models are back online after a two-week standoff with the federal government, settling the company and the administration into what observers describe as a fragile truce. The export-control action was triggered after Amazon flagged concerns that safeguards on Fable could be circumvented; Anthropic says it has since built a fix, backed by the Commerce Department's Center for AI Standards and Innovation. The exact terms Anthropic agreed to in order to secure the reversal remain undisclosed.
The episode, alongside OpenAI's own government-requested deferral of GPT-5.6's public rollout, underscores a broader shift: regulators are now treating frontier model access itself as a national-security lever, not just a product decision.
↑ back to top
19 NEW STATE AI LAWS ENACTED IN TWO-WEEK WINDOW ENDING LATE JUNE
Policy Desk | Source: AI Governance Institute
Nineteen new AI-related laws were enacted across eleven states and Congress in a single two-week window ending in late June 2026, adding to an increasingly layered state-by-state compliance environment for AI developers. The state-level activity runs alongside a separate June 2 federal executive order, "Promoting Advanced Artificial Intelligence Innovation and Security," which directs a forthcoming (for now voluntary) frontier-model benchmarking process rather than a mandatory licensing regime.
↑ back to top