Anthropic Launches Fable 5.2 and Mythos 5.2
Plus, π How to keep the same brand character in every image with ChatGPT, Meta launches Muse Ember, a low-cost AI model built for everyday agent tasks, and more!


Hola Decoderπ
If someone forwarded this to you and you want to Decode the power of AI and be limitless, then subscribe now and Join Decode alongside 30k+ code-breakers untangling AI.

π§ Anthropic launches Claude Fable 5.2 and Mythos 5.2 with major coding and reasoning upgrades
Anthropic has introduced Claude Fable 5.2 and Mythos 5.2, its latest AI models designed to handle complex engineering, scientific research, and long-running tasks, with improvements in coding, reasoning, vision, and response speed.

The Shift:
- Claude Fable 5.2 can coordinate teams of AI agents and tackle complex software engineering projects, including large-scale code migrations and technical tasks that previously required extensive human involvement.
- Anthropic says its models helped solve two open problems in black hole physics, design three working electronic devices, and formalize a proof of Fermat's Last Theorem in 30 hours during prerelease testing.
- Fable 5.2 delivers responses 30% faster than its predecessor, while its input pricing drops from $10 to $8 per million tokens. The model also supports zero-data-retention policies for enterprise customers.
- Fable 5.2 is available now, while Mythos 5.2 is initially restricted to vetted organizations, as Anthropic continues evaluating advanced model capabilities and potential safety risks.
Anthropic is pushing Claude beyond everyday AI assistance toward complex scientific and engineering work. The advances could make AI agents more useful for long-term projects, although questions remain about how reliably these capabilities can be evaluated and deployed.

Together with New Engen
Your next winning ad needs a different voice

Your product solves a real problem. But when every ad makes a promise, buyers need a reason to believe yours.
A qualified expert explaining how it works can give them that reason.
Grapevine, now part of New Engen, creates ads delivered through creator and publisher handles with production managed from brief to launch.
You approve the angles and ship product. Grapevine sources creators, handles briefs and payments, then produces and posts the content.
- Madison Reed achieved 20% efficiencies over target CPA and 50% higher LTV.
- Mathnasium cut Meta CPL by 33% in under 30 days.
- Cirkul achieved 15-50% lower TikTok CPA and nearly doubled CTR through creator whitelisting.
Put this approach to work with your first campaign on Grapevine.
Kick off by November 30 and your first campaign is free: 8 statics and 1 article worth $1,500.

π How to keep the same brand character in every image with ChatGPT
ChatGPT Images 2.6 launched on November 4 on every plan. You can now save a reference image under a name and call it back in any chat, so your mascot looks identical in every post.

1. Approve one master image - Generate or upload your character: full body, facing forward, plain background. This becomes the source of truth.
2. Save it by name - βSave this image as a reference named [Pip] so I can reuse this exact character in future chats.β
3. Write the identity rules - Keep a short rule line to paste with every request: β[Pip] always has [colors, outfit, proportions]. Never change these.β
4. Batch a month of content - In a new chat: βUsing [Pip], create four images: a launch post, a sale banner, an FAQ graphic and a thank-you post. Change only the scene and pose.β
5. Add motion - Turn the best one into a looping GIF for your next email.
Only save photos of real people, like customers or teammates, with their permission.

π€ Meta launches Muse Ember, a low-cost AI model built for everyday agent tasks
Meta has introduced Muse Ember, its smallest and most cost-efficient AI model, designed to handle repetitive tasks such as coding, tool use, and document analysis while keeping the cost of running AI agents low.

The Shift:
- Muse Ember costs just $0.10 per million input tokens and $0.50 per million output tokens, making it a lower-cost option for businesses running AI agents at scale.
- Meta says Ember matches Claude Haiku 5.5's top intelligence score at roughly one-third the cost, while outperforming GPT-6 Luna across its tested reasoning settings on the Artificial Analysis benchmark.
- The model can process text, images, audio, and video, with a 1-million-token context window, allowing agents to analyze lengthy documents, interpret screenshots, and work with recordings without switching models.
- Ember scored 37.8% on AutomationBench, compared with 20.2% for GPT-6 Luna and 13.9% for Claude Haiku 5.5, and is designed to work alongside Muse Wonder, which handles more complex planning.
Meta is focusing on making AI agents cheaper to operate without giving up useful capabilities. By pairing powerful models for planning with smaller ones for repeated tasks, businesses could automate more work at a lower cost.

Together with Playbook Pro
Your Next 6 Months of Ad Spend Could Disappear in 6 Weeks

CAC is up 60% this year, and 70% of first-time buyers never return. Keep leaning on ads, and the budget youβre planning for the next half year could be gone before spring.
Rewriting the DTC Rulebook is the shortcut top operators are using to escape the trap, a 24-page report distilled from 200+ hours of real brand research and field tests.
Inside, youβll learn how to:
π₯ Build emotional campaigns that double repeat-purchase rates in 90 days.
π° Raise prices without losing volume using the Brand Value Canvas.
π§ Cut 30% of paid-ad dependency with the Owned Media Checklist.
Download today and get 2,000+ viral video hooks already tested to grab attention in under 3 seconds, ready to plug into your next launch.
Stop feeding the ad machine.
Grab the free report now and lock in sustainable growth before your next dollar disappears!

π Tools you Cannot Miss:
π¬ ClipsPal β Automatically removes failed takes and silences from recorded videos, adds captions, zooms, and visuals, and repurposes footage into short clips and articles.
π± skuttr β Turns a one-sentence description into an Instagram Reel by finding relevant photos and videos in your iPhone camera roll and arranging them into a vertical video.
π Watchflows β Builds visual Mac automations from plain-language instructions, handling tasks such as organizing screenshots, sorting downloads, and processing documents with AI.
π₯ Cardboard AI β Turns prompts and uploaded footage into editable videos, generating motion graphics, sourcing B-roll, adding captions, and refining cuts through natural-language instructions.
π Dageno AI 2.0 β Tracks which brands AI search engines recommend, identifies competitors and citation sources, and suggests content improvements to increase brand visibility.

π Quick Hits
π¬ Paramount Skydance CEO David Ellison called AI potentially the biggest creative breakthrough in modern history, predicting a greater impact on filmmaking than CGI while emphasizing it will never replace human artists.
ποΈ Republican House committee chairs requested a closed-door briefing with Nvidia CEO Jensen Huang on November 18, seeking details on security protocols against foreign espionage and his reported interactions with Alice Huang.
π Apple introduced $49.99 Intelligence Usage Packs in Logic Pro, allowing users to purchase additional access to AI-powered music creation features after reaching their included usage limits.
π¨ Instagram launched AI-powered Effect Builder, allowing users to create custom Stories filters using text prompts and reference images, with the web-based tool currently available to US residents aged 18 and older.
π Otter.ai is expanding transcription support to 31 languages, adding 27 new languages, with upcoming AI-powered detection that automatically follows speakers switching languages during meetings without requiring manual settings changes.
π¬ Disney and Adobe announced a multiyear partnership to develop custom, artist-led generative AI models using Firefly Foundry, supporting production workflows across Disney Studios, theme parks, cruise experiences and ESPN content.

π§© Prompt of the Day

Turn Any Photo Into a Playful Hand-Drawn Doodle Artwork
Image Editing Prompt: Transform the uploaded photograph into a creative photo-doodle artwork by adding playful, hand-drawn illustrations that interact naturally with the original subject while preserving the photograph's identity, composition, and lighting.
Context
- Reference Image: [Upload your photo]
- Style: Creative hand-drawn photo doodles
- Doodle Aesthetic: Playful, colorful, spontaneous illustrations
- Interaction: Doodles responding to the subject's pose, movement, or surroundings
- Colors: [Bright multicolored / Pastel / White line art]
- Aspect Ratio: Match the original image
Instructions
Analyze the uploaded photograph and identify the main subject, its pose, gestures, silhouette, and surrounding environment. Preserve the original subject's appearance, identity, proportions, clothing, background, composition, and lighting without modification. Add imaginative hand-drawn doodles that interact directly with the subject, following body contours, extending gestures, emphasizing movement, or responding creatively to nearby objects. Incorporate playful elements such as expressive outlines, motion trails, stars, flowers, arrows, swirls, tiny characters, or whimsical shapes wherever they naturally complement the scene. Make the doodles feel spontaneously drawn over the photograph with slightly imperfect strokes, varied line thickness, and balanced placement. Keep the subject clearly visible and avoid overcrowding the image.

π€³AI Nugget of the Day


Thanks for Decoding with usπ₯³
Your feedback is the key to our code! Help us elevate your Decode experience by hitting reply and sharing your input on our content and style.
Keep deciphering the AI enigma, and we'll be back with more coded mysteries unraveled just for you!
