Everything Claude Launched This Month (in 13 minutes)
What this video is
⚡ a 13-minute video, readable in 60 secondsAnthropic released a new model family this month, Opus 5.5, Sonnet 5.5, and Haiku 5.5, with Opus 5.5 costing about 40% less than the previous Opus (prior price not stated). The bigger structural delta is that Cowork and chat have merged into one Claude; previously chat handled quick questions and Cowork handled bigger multi-step work, and work did not carry over between the two. Alongside the merge, Anthropic launched five new in-chat tools (Docs, Slides, Design, Dashboards, and Motion), added Claude to Google Workspace and Apple CarPlay, and shipped three Claude Code updates (redesigned projects, a wrap-up usage allowance, and mods). For anyone tracking Claude's roadmap, this collapses previously separate tools and workflows into a single interface and extends Claude into documents, data, video, and cars.
New: Anthropic released a new model family this month, Opus 5.5, Sonnet 5.5, and Haiku 5.5 [00:57]
Key takeaways
- New: Opus 5.5 is the biggest and smartest model, built for long hard jobs like deep research [01:09]
- Repriced: Opus 5.5 costs about 40% less than the previous Opus; prior price not stated [01:18]
- New: Sonnet 5.5 is the everyday model, faster and cheaper than Opus, for documents, slides, spreadsheets, and coding [01:27]
- New: Haiku 5.5 is the cheapest and fastest model Anthropic has ever made [01:39]
+ 30 more takeaways
- Changed: Cowork and chat are now merged into one Claude [02:23]
- From -> To: previously chat handled quick questions and Cowork handled bigger multi-step work as two separate places -> now merged into one Claude [02:32]
- Fixed: previously work did not carry over between chat and Cowork [02:39]
- Fixed: anyone who used Cowork before had their chats, projects, connectors, and skills carry over automatically [03:15]
- Changed: by default Claude now asks before taking an action, but users can switch it to only check in when needed [03:11]
- New: Anthropic shipped three new tools, Docs, Slides, and Design, used just by asking Claude in a normal chat [03:30]
- New: created documents, decks, or designs are saved in the Artifacts tab [03:45]
- New: Docs are private until shared and can be exported to Word, PDF, or Google Docs [04:01]
- New: Slides can be edited manually or by asking Claude, presented from inside Claude, or downloaded as PowerPoint or PDF [04:21]
- Changed: Docs, Slides, and Design are now out of beta and work on every plan, including the free plan [04:42]
- New: Dashboards let users describe what they want in plain words, and Claude writes the queries, builds the charts, and keeps them updated [05:20]
- New: Dashboards work with Snowflake, BigQuery, and Salesforce [05:27]
- New: clicking any number in a dashboard shows where it came from, and every chart shows when it last refreshed [05:32]
- New: dashboards can be sent to other tools like Mixpanel, PostHog, or Grafana [05:39]
- Availability: Dashboards are on paid plans [05:39]
- New: Motion, in beta, makes animated videos from a single prompt by writing code that animates your own words, numbers, charts, shapes, and images, with no AI video generator or fake footage [06:01]
- New: Motion videos can be exported as a file or opened in tools like Descript or Runway, started by typing /motion in a chat [06:37]
- Availability: Motion is only available on team and enterprise plans [06:37]
- New: Claude for Google Workspace puts Claude inside Google Docs, Sheets, and Slides via a sidebar from the Google Workspace Marketplace, asking permission before each edit [07:02]
- Availability: the Google Workspace integration is in beta on Pro, Max, Team, and Enterprise plans [08:02]
- New: Claude was added to Apple CarPlay, letting users open Claude from the CarPlay screen and talk hands-free [08:15]
- Unchanged limit: CarPlay Claude can't control the car or phone, and has no wake word, so the app must be opened each time [08:43]
- Changed: Claude Code is getting a redesigned 'projects' feature, in beta, built around a goal rather than just a folder of chats [08:57]
- New: a main Claude splits a goal into smaller jobs that run in parallel threads with shared memory for decisions [09:14]
- Availability: the redesigned projects feature is now available to every Pro and Max user [09:52]
- New: a soft-landing usage allowance gives Claude extra usage to reach a stopping point if the usage limit hits mid-response, shown as a 'Wrapping Up' message; that extra usage still counts toward the weekly limit and won't always be enough to finish the job [10:40]
- New: Mods are small add-ons that change how Claude Code looks or behaves, described as like a browser extension but for Claude Code [11:02]
- New: no coding is needed to make a mod, users can ask Claude Code to build one, approve it, and turn it on for that session [12:20]
- Caution: mods have the same access to your computer as Claude Code, so only install ones you trust [12:34]
- Why it matters: the presenter's recommended first thing to try this week is asking Claude to build a deck from your notes [13:12]
![New: clicking any number in a dashboard shows where it came from, and every chart shows when it last refreshed [05:32]](assets/imkKVWoIzVM/frame_333.jpg)
Shown on screen — grab and go
CODEClaude Motion animate() API call
animate({ words: "Fernhill" numbers: 4200, charts: line(rides), shapes: morph(shape), images: slides(3), [...]merged from OCR reads at 6:21-6:25; trailing closing syntax is garbled and marked with [...] shown at 6:25
PROMPTIn-Docs sidebar editing instruction
Tighten up the executive summary and lead with recommendation. Then clean up the formatting with clean headings and put the next steps into a table.
shown fully legible in the Claude for Docs sidebar box, confirmed across 7:31-7:37 shown at 7:31
PROMPTLaunch-deck request typed in chat
here's the launch kit for the Varde 2.0. build the reveal deck for fri's readiness review.
reconstructed from crushed-space OCR at 3:42, spaces restored only shown at 3:42
CODEClaude Code .claude/agents subagent roster
PROJECT .claude/agents · 7 code-reviewer sonnet run Reviews the current diff for correctness, bugs, risky patterns [...] dependency-doctor sonnet run Checks package manifests for outdated, deprecated or vulnerabl[...] docs-writer haiku run Keeps README and inline docs in sync with the code. Rewrites o[...] perf-profiler sonnet run Looks for the slow parts. Spots N+1 queries, repeated work inside[...] release-notes haiku running Turns the commits since the last tag into user-facing release note[s] security-auditor opus run Read-only pass over the codebase for injection, auth and secrets[...] test-writer sonnet run Finds code paths without test coverage in the files you changed a[...]
merged duplicate OCR passes at 12:08 and 12:12; most descriptions are cut off mid-sentence, marked with [...] shown at 12:08
PROMPTCore Web Vitals goal given to Claude
Our Core Web Vitals for the marketing site have been dropping for a month: p75 LCP is 4.1s on mobile and CLS is 0.28, [...]
transcribed from the screen at 9:14; sentence is cut off at the end, marked with [...] shown at 9:14
How this brief was shaped: What Changed · confidence Medium
Transcript opens with a creator stating he reviewed every Anthropic release this month and is boiling down 10 new things, then promises for each launch what it is, why it is useful, and whether it is worth your time. That is a consumer-facing roundup of new releases rather than raw technical release notes, so it fits the leads-with-whats-new shape over a github-style changelog.
The lens sets this brief's structure, never its facts — every claim is held to the same citation and fact-check standard.
Jump to a moment
Their links, sorted & clickable
🏛️ Communities & courses1
👉Was this brief useful?