Anthrogen Unveils Odyssey Protein Model

Welcome to another edition of AI PlanetX.

Anthrogen unveils 102B-parameter protein model; DeepMind CEO calls GPT-5 math claims “embarrassing”; novel prompting method boosts LLM creativity.

• Hottest AI News

• Top AI & SaaS Tools

• AI Tutorial: Replace Voice in AI Videos

• Top AI & Tech News

• AI Art Spotlight

• Prompt of the Day: Ultimate YouTube Script Prompt

• AI Video Tutorial

• Weekly Creator Playbooks to Grow on Social

Hottest AI News

Anthrogen

Anthrogen Unveils Odyssey, the World's Largest 102B Parameter Protein Language Model

Anthrogen has unveiled Odyssey, a 102-billion-parameter protein language model — the largest and most advanced to date. It marks a major leap in molecular design, bringing us closer to engineering proteins with the same precision used to design machines.

Details:

• Odyssey is a multimodal protein model that learns from sequence, 3D structure, and function. It supports generation, editing, and co-design across 1.2B–102B parameters. The model encodes amino acid sequences and 3D shapes into compact tokens with added cues like domains and structure hints

• The backbone introduces Consensus blocks, which replace self-attention with a method that scales linearly with sequence length. Instead of all-to-all attention, Consensus lets nearby regions agree locally before spreading outward

• Training uses discrete diffusion, adding noise like mutations and learning to restore correct sequences and structures. This outperforms masked modeling and teaches coordinated multi-residue updates that keep geometry and sequence aligned

Odyssey demonstrates remarkable data efficiency, achieving better performance than competing models with approximately 10x less data—crucial given protein data scarcity.

Stay up-to-date with AI

The Rundown is the most trusted AI newsletter in the world, with 1,000,000+ readers and exclusive interviews with AI leaders like Mark Zuckerberg, Demis Hassibis, Mustafa Suleyman, and more.

Their expert research team spends all day learning what’s new in AI and talking with industry experts, then distills the most important developments into one free email every morning.

Plus, complete the quiz after signing up and they’ll recommend the best AI tools, guides, and courses – tailored to your needs.

Sign up to start learning.

OpenAI

GPT-5 Math Claims Were "Embarrassing" Says DeepMind CEO

OpenAI researchers recently claimed GPT-5 had solved long-standing math problems, a breakthrough that drew huge attention. But the claim soon unraveled, exposing misinterpretation and overhyped results that embarrassed the company.

Details:

• OpenAI manager Kevin Weil sparked controversy with a deleted tweet claiming GPT-5 had solved “10 previously unsolved Erdős problems.” Other researchers echoed the claim, suggesting the AI had cracked decades-old challenges and opened fresh paths in mathematics

• The story fell apart when mathematician Thomas Bloom clarified that “open” on his site just meant he didn’t know the answers—others had already solved them. GPT-5 had simply resurfaced existing research. DeepMind’s Demis Hassabis called it “embarrassing,” Meta’s Yann LeCun said OpenAI had “bought into its own hype”

• Still, GPT-5 showed real value as a literature review tool, finding scattered papers and unifying results hidden by inconsistent terminology. Mathematician Terence Tao noted that this is AI’s real strength: not solving the hardest problems, but speeding up research

OpenAI’s pressured, sloppy messaging raises doubts about why top researchers would publicize dramatic, unverified claims in a hype-filled, billion‑dollar field.

Top AI & SaaS Tools

• FormRobin (Life-time Deal): Create professional, AI-generated forms to capture leads, run surveys, and collect feedback without branding or hidden upsells

• Kiro: Amazon launched Kiro, a Claude-powered AI dev tool built on VS Code, rivaling Windsurf and GitHub Copilot [F-R-E-E]

• YouMind: Helps anyone capture ideas and create polished articles, podcasts, videos, and other content [F-R-E-E]

• Supamail App: Receive a clean daily digest summarizing the last 24 hours of emails, organizing by sender and category, and letting you mute unwanted alerts [F-R-E-E Trial]

• Manus 1.5: Latest update adds unlimited context and tools to build full-stack AI web apps, including backends, user logins, custom domains, and analytics [F-R-E-E]

The AI Insights Every Decision Maker Needs

You control budgets, manage pipelines, and make decisions, but you still have trouble keeping up with everything going on in AI. If that sounds like you, don’t worry, you’re not alone – and The Deep View is here to help.

This free, 5-minute-long daily newsletter covers everything you need to know about AI. The biggest developments, the most pressing issues, and how companies from Google and Meta to the hottest startups are using it to reshape their businesses… it’s all broken down for you each and every morning into easy-to-digest snippets.

If you want to up your AI knowledge and stay on the forefront of the industry, you can subscribe to The Deep View right here (it’s free!). 

AI Tutorial

How to Replace Any Voice in Your AI Videos

Tired of every AI video sounding robotic? With a few simple tools, you can replace any default AI voice with a custom one that matches your style — from cinematic narrations to branded voiceovers.

1. Generate AI Video

   Use Google Veo (or any text-to-video generator like Pika or Runway) to create your video. Once you’re happy with the visuals, download the MP4 file.

2. Check Audio Quality

   Before replacing the voice, play your video and note:

   • Length of the clip

   • Type of voice or background sound

   This ensures your replacement audio will sync properly later.

3. Extract Original Audio

   Use a video-to-audio converter (e.g., Media io, Convertio, or VEED) to extract the MP3 track from your video.

   Save the file with a clear name like “scene1_original.mp3”.

4. Clean and Prepare Audio

   (Optional but recommended) Run the MP3 through a noise reducer such as Auphonic or Audacity. This removes unwanted background noise before cloning or replacing.

5. Clone or Choose Voice

   Go to ElevenLabs Voice Changer.

   You can either:

   • Use one of their pre-made professional voices, or

   • Clone your own voice using a 1-minute sample.

   Upload your audio and let ElevenLabs generate your custom voice version.

6. Adjust Voice Settings

   Tweak these settings for best results:

   • Stability: Controls tone consistency

   • Similarity: How close it stays to the chosen voice

   • Style exaggeration: Adds emotion or energy

   Experiment with small adjustments to match the mood of your video.

7. Download Audio

   Export your generated audio as MP3 or WAV. Rename it clearly (e.g., “scene1_customvoice.mp3”) so it’s easy to align later.

8. Sync Audio and Video

   Open CapCut (or Premiere Pro/DaVinci Resolve).

   • Import the original MP4 video

   • Add the audio layer

   • Mute original track

   • Align the voice using the waveform for lip-sync precision.

Note: Record a few seconds of silence at the start of your sample audio before cloning. It helps AI tools capture cleaner background noise profiles.

Top AI & Tech News

• Researchers discovered a simple prompting method—Verbalized Sampling (VS)—that sharply boosts creativity and output diversity in LLMs

• Google added live Maps grounding to Gemini, letting developers pair its reasoning with real-time and historical Maps data for location‑aware, factual responses

• Codev is an open-source platform and methodology that turns “vibe coding” conversations into structured, versioned, auditable parts of the codebase

• Wikipedia’s human pageviews dropped about 8% year over year partly due to AI search summaries and users turning to social video for answers

AI Art Spotlight

Create Your Own Photo Like This Using the Prompt Below:

Use the attached image as the subject. Preserve my face. Full-body portrait in Captain-style tactical suit with star chest plate, red white accents, navy panels. No helmet, hair visible. Hold a round vibranium shield at side or on forearm. Strong heroic stance, warm golden hour light. 8k photoreal, entire body in frame.

Negative prompt
no helmet covering face, no cartoon, no grunge dirt on face, no cropped feet, no bent shield, no text

Output settings
photo style, 4:5, 4096 x 5120 or 7680 x 9600, fabric weave and stitching visible, identity match high

Model: Gemini 2.5 Flash Image (Nano Banana)

Prompt of the Day

Viral YouTube Prompt That Hooks Viewers

This mega prompt turns AI into an expert YouTube scriptwriter modeled on MrBeast, MKBHD, and Ali Abdaal. It guides you to craft high-retention videos with a laser-focused hook, relentless pattern interrupts, and strategic open loops.

It enforces tight pacing, active voice, and a clear story arc from problem to transformation, while embedding visual notes, B-roll cues, text overlays, and sound effects for cinematic impact.

Top AI Video Tutorial

7 Mind-Blowing Use Cases for Perplexity Comet Agents (Browser)

Want Weekly Creator Playbooks to Grow on Social?

Want to stand out and grow your social presence? Join 10K+ creators getting weekly playbooks to build a creator business that pays.

Unlock $10,000+ Worth of F-R-E-E Creator Blueprints, including: 👇