Free Claude Startup Perks Paused After a 21x Surge
Anthropic paused its free Claude startup perks just 48 hours after launch because applications jumped 21x, according to Forbes. No restart date has been reported.
The broader Claude for Startups program is still live. It was expanded on October 6 with a $45K partner stack and $1K in API credits, per CNBC.
Cyber Mission Launches With a Free Scanner for Open Source
Anthropic launched a Cyber Mission on October 8, including a critical infrastructure program with 11 founding partners. A day later it released a Claude-powered scanner that offers free vulnerability audits to open-source maintainers.
It follows the October 6 expansion of the Cyber Verification Program, which now has three access tiers and folds in Project Glasswing.
Sonnet 5.5 Cache Reads Drop to $0.10 Per Million
Claude Code 2.1.296 prices Sonnet 5.5 cache reads at $0.10 per million tokens, down from $0.20. If your agents re-read big contexts, that is a straight 50% cut on that line item.
The release also lifts the MCP tool description and server instruction limit from 2,048 to 4,096 characters, adds a Read tool allow_large option, and adds a CLAUDE_CODE_WORKFLOW_SUBAGENT_MODEL variable to run all workflow agents on one model.
Google Workspace Sidebar and Claude Code Mods
Claude for Google Workspace entered public beta on October 6, adding a sidebar for Docs, Sheets and Slides plus Google file connectors.
Mods for Claude Code launched October 1 in the CLI and desktop app. They let you add TypeScript functions that change prompts, UI and built-in features.
Reports: Anthropic Lines Up SpaceX and IREN Compute
The Wall Street Journal reports co-founder Tom Brown brokered a compute deal with SpaceX worth about $1.25B a month. The Information separately reports an unannounced cloud deal with IREN, where a 245 MW site could generate about $1.3B a month.
Neither deal has been confirmed by Anthropic. Treat the numbers as reported, not official.
Palantir CTO Downplays the Pentagon Split
On Face the Nation, Palantir CTO Shyam Sankar disputed a Bloomberg report about an Iran strike and downplayed the Pentagon’s decision to stop using Anthropic, per CBS News.
Elsewhere, Wired reports HarperCollins, Simon & Schuster and Hachette have quietly used AI tools including Claude to write back-cover copy and pitches.
Capacity Is the Moat
Two reported deals worth over a billion dollars a month each, a free perk paused after 48 hours, and a cache-price cut all point the same way. Demand is outrunning supply, and Anthropic is racing to buy power while nudging heavy users toward cheaper patterns.
Pair that with the White House pushing for immediate AI incident disclosure and last week’s offline evals, and the pitch is clear: more compute, tighter guardrails. Whoever scales both wins.