Skip to content
Back to The Briefing

Claude and Claude Code: Fable 5.2 brings higher scores and lower token prices

The Briefing··2 min read

This morning’s briefing leads with Claude Fable 5.2, which brings a substantially higher independent benchmark score and lower API token prices than its predecessor. Copilot can now change information in connected workplace services, while Google has restarted its delayed Workspace Skills rollout.

New releases

Claude and Claude Code. Fable 5.2 arrives on paid Claude plans, subject to workspace controls, with a Max-effort Artificial Analysis Intelligence Index score of 65 versus Fable 5.1’s 49 (about 33% higher), roughly double the benchmark cost per task (US$26.54 versus US$12.99), 20% lower standard API prices of US$8 input and US$40 output per million tokens, and temporary inclusion for Pro and Team Standard seats up to 50% of weekly usage through 07:59 GMT on 17 November. More

Microsoft Copilot Chat and Cowork. Microsoft 365 Copilot licence holders can now create, update or delete information through supported, administrator-enabled connectors in Chat and Cowork, including Jira, Notion and Canva, with approval requested by default; write actions in Office apps are still forthcoming. More

Gemini in Google Workspace. English-language Skills are rolling out to eligible Workspace users aged 18 and over, bringing reusable, stackable instructions to Gmail, Docs, Slides, Drive, Chat and Studio, with Rapid Release starting on 9 November and Scheduled Release on 30 November, each taking up to 15 days. More

Google Workspace Studio. The new Wait until step pauses an automated workflow for up to 14 days before continuing, with rollout to eligible Business, Enterprise and Education accounts starting on 9 November and taking one to three days. More

Adobe Firefly. November’s update adds text editing inside images and desktop relighting with up to three light sources, while AI Assistant beta users gain reusable brand guidance, plan approval and creative memory that is on by default and rolling out in stages. More

ElevenLabs ElevenAgents. ElevenAgents users can now run outbound campaigns across phone calls and WhatsApp with automatic retries, callbacks and context carried between conversations, replacing batch calling; workspaces and agents using Zero Retention Mode are excluded. More

Elsewhere in AI

Australia plans an AI safety regulator, mandatory incident reporting and possible criminal penalties following the OpenAI Medicare breach, The Guardian reports. Read

Palantir lobbied a health minister to extend its AI tools to NHS GP records, which its existing contract excludes, The Guardian reports. Read

Qantas plans to cut 850 Australian jobs and outsource work to India under an AI-driven overhaul involving Accenture, The Guardian reports. Read

Get The Distillation