Good Morning! Here's what I have for you in today's newsletter:
Anthropic releases Claude Sonnet 5.5, faster and cheaper
ElevenLabs launches Eleven v4, voice clones from just 10 seconds of audio
Manus 2.0 adds a video timeline, game builder, and more
Build a free WhatsApp assistant with Opus 5.5 and OpenClaw
4 new AI tools worth trying today
AI MODELS
Anthropic released Claude Sonnet 5.5, the second model in its Claude 5.5 family, keeping Sonnet 5's price of $2 per million input tokens and $10 per million output tokens.
Sonnet 5.5 runs more than 30 percent faster and uses fewer tokens, cutting cost per task by up to 30 percent. At Low or Medium effort, it beats Sonnet 5's best scores for about a tenth of the cost.
The model scores 70.6 percent on Terminal-Bench 4.0, above Opus 5.5's 66.4, while trailing it slightly on CursorBench 4.0 at 55.5 percent against 57.8.
It is the first Sonnet with cyber safeguards, routing high-risk security requests back to Sonnet 5, and it is available in Claude apps, the API, AWS, Google Cloud, and Azure.
If you've been paying for Opus to write code or build documents, Sonnet 5.5 can likely handle much of that work for half the price. It won't match Opus on every task, so try it on a few of your real jobs before you switch.
AI MODELS
ElevenLabs released Eleven v4 and Eleven v4 Turbo, two speech models on a new architecture that rank first on Artificial Analysis, with Turbo running at about 100 ms median latency.
Instant Voice Clones now capture a voice from 10 seconds of audio. Long-form narration and dialogue hold the same speaker identity across regenerated lines.
Inline tags direct emotion, pacing, reactions, sound effects, style, or a plain description of the delivery, with IPA support for exact pronunciation.
Both models cover 90+ languages, including Cantonese, Mongolian, and Odia. For the next two weeks, the API is discounted to $22 per million characters for v4 and $11 for Turbo.
You only need 10 seconds of your voice to get a clone that stays consistent across a full audiobook or a long podcast. Turbo answers fast enough that a support agent built on it feels like a real conversation. If you're on a Creator plan or higher, v4 is free in ElevenCreative for the next two weeks.
AI AGENTS
Manus launched Manus 2.0 on web, desktop, and mobile, built on Cascade, a new agent framework that used 23 percent fewer tokens and cost 32 percent less to run in testing.
The desktop app, now called Manus Studio, opens generated videos on an editable timeline. Music, clips, and images can be replaced individually, with the rest of the video left unchanged.
A new Game Dev workspace turns a prompt into a playable game, with tools to edit its art, code, and scenes. Finished games publish to a shareable link, and multiplayer runs on a paid Cloud Computer.
Automations start tasks from events in connected services, such as a new email, a Slack message, or a Notion update, set up in a single prompt.
You can now fix one detail in an AI-generated video without generating the whole thing again. A launch ad that's nearly finished can be adjusted on the timeline in minutes, with your own footage or a different track. Game Dev works the same way, with art and code editable after the first version.
HOW TO AI
A self-hosted agent on your own Claude subscription, no separate API bill.
Before you start, confirm Node.js 24.16+ (below 25) or 26.1+ (node --version, OpenClaw's own requirement) and Claude Code (npm install -g @anthropic-ai/claude-code), since Step 2's token comes from its CLI.
Step 1: Install and onboard
npm install -g openclaw@latest
openclaw onboard --install-daemon--install-daemon keeps it running as a background service after you close the terminal.
Step 2: Skip the API bill
Type these directly into your own terminal, not through OpenClaw's own interface, they need an interactive terminal session:
claude setup-token
openclaw models auth setup-token --provider anthropic
openclaw models statusIf the second command wants a TTY it doesn't have, use openclaw models auth paste-token --provider anthropic instead.
This generates a subscription-scoped token from your Pro or Max plan instead of a pay-per-token API key.
Step 3: Link WhatsApp
openclaw channels login --channel whatsappScan the terminal's QR code from WhatsApp: Settings, then Linked Devices, then Link a Device.
If you use iPhone Mirroring, turn it on right after this step, not before, mirroring disables the camera so it can't do the QR scan itself.
Step 4: Control who can reach it
Fastest: your own number, with selfChatMode enabled, no new SIM needed. Cleaner separation: a spare phone with an eSIM and WhatsApp Business.
openclaw config set channels.whatsapp.dmPolicy pairing
openclaw pairing approve whatsapp <code>Step 5: Give it a job
Every morning at 7, look at my calendar and any unread messages from the last 12 hours, and text me a short summary: what's on today, what needs a reply, and anything urgent. Don't act on anything, just report.
P.S. You can access all the AI courses and trainings (including the full version of this one), prompts and workflows if you upgrade.
Kling opened Kling 4.0 Flash to Ultra Yearly subscribers ahead of the full Kling 4.0 release in October, which adds 4K and 10-bit HDR output, up to 15 multimodal references, 10 keyframes, and native 30-second generation.
Perplexity added Profiles, Skills, and managed connectors to its Agent API. A developer can now configure an agent's model, instructions, and tools once under a versioned ID and reuse it across apps and team projects.
Cua released Cua Perception, which lets its open-source Driver read pixels on canvases, games, and remote desktops, turning a screenshot into labeled regions locally on the CPU so any model can pick a target instead of guessing coordinates.
🎬 Krea: added Seedance 2.5 Draft Mode, generating test clips at 480p before switching to 1080p once a scene is right.
🎧 Fish Audio: its speech-to-text model now identifies speakers and labels emotion cues like [laughter] and [surprised] across 83 languages.
📱 Rork: added iPhone Duo support, so apps can be built for both screens and tested on a live Duo in a 3D simulator.
🤖 Devin: opened a beta of Devin Mobile, with a waitlist to run and check on Devin sessions from a phone.
THAT’S IT FOR TODAY
Thanks for making it to the end! I put my heart into every email I send, I hope you are enjoying it. Let me know your thoughts so I can make the next one even better!
See you tomorrow :)
- Dr. Alvaro Cintas
✓ Full archive of premium guides with ready-to-use prompts
✓ Structured AI courses (step-by-step, start-to-finish)
✓ Every upcoming premium tutorial
✓ $500+ worth of AI tools