Barclays Scales Claude Across the Bank
Anthropic and Barclays expanded their work together on October 1. More than 16,000 colleagues already use a Claude-powered Colleague Knowledge Assistant, which handles over one million searches to support 20+ million UK retail customers.
In Global Markets, Claude classifies and routes about 120,000 emails a day. Barclays expects Claude Code to reach 50% of its developers by the end of 2026.
Sonnet 5.5 Rounds Out the 5.5 Family
Claude Sonnet 5.5 launched September 28 as a faster, lower-cost sibling to Opus 5.5. It runs 30%+ faster, costs up to 30% less per task, and scores 70.6% on Terminal-Bench 4.0. Per-token pricing holds at $2 input and $10 output per million.
It is the first Sonnet with Opus-level cybersecurity safeguards, and it is on Claude Platform, AWS, Google Cloud and Azure.
Claude Code 2.1.287 Introduces Claude Mods
The October 1 release lets plugins modify deeper Claude Code behavior, and ships a built-in “You should know” mod. MCP servers can now prompt users with URLs for browser-based auth flows under the 2025-11-25 protocol.
Reliability got love too: Remote Control reconnects with a 30-second timeout and retry, large MCP tool results use less memory, and Windows Bash is faster.
Migrating to Sonnet 5.5? Change the Thinking Setting
Moving to claude-sonnet-5-5 means switching the thinking setting from off to between_tools. Old calls that assumed the previous behavior will need a tweak.
Cache reads run $0.20 per million tokens and cache writes $2.50. Zero data retention is available.
Anthropic Reportedly Sets a Pre-IPO Investor Day for Oct. 14
Bloomberg reports Anthropic plans investor meetings on October 14, with an IPO as early as mid-November and marketing possibly starting the week of November 9. Valuation talk runs from $1.8 trillion to $2 trillion.
Reports put 2025 revenue at about $4.6 billion, up 12x. These are reported plans, not announced ones.
Appeals Court Upholds Pentagon’s Supply-Chain-Risk Label
In a 2-1 ruling on September 25, the D.C. Circuit backed the Pentagon’s March designation of Anthropic as a supply chain risk. The fight centers on Anthropic refusing to drop limits on lethal autonomous warfare and mass surveillance of Americans.
Judge Henderson dissented, warning the reading could enable coercive demands on future contractors.
Anthropic Is Selling Trust to Banks, Governments and Soon Public Markets
Barclays, last week’s FedRAMP High launch and an IPO roadshow all point the same way: Anthropic wants to be the safe, regulated-grade AI vendor. Big, sticky enterprise deals are exactly what public investors will want to see.
The tension is the Pentagon. A court just upheld a label that sits awkwardly beside that pitch, and it will show up in the risk factors. Watch October 14.