- AI Valley
- Posts
- SeeDance 2.0 hype reaches new heights
SeeDance 2.0 hype reaches new heights
Plus: Gemini 3 Deep Think is now world’s top-performing model
Together with
Howdy, it’s Barsee again.
Happy Friday, AI family, and welcome back to AI Valley.
Today’s climb through the Valley reveals:
Anthropic has raised $30B in funding at a $380B valuation
Spotify says top developers haven’t coded since December due to AI
Gemini 3 Deep Think is now world’s top-performing model
OpenAI released GPT-5.3-Codex-Spark, a real-time coding model
SeeDance 2.0 hype reaches new heights
Plus trending AI tools, posts, and resources
Let’s dive into the Valley of AI…
FLOW
Wispr Flow turns "thinking out loud" into crisp, structured text you can paste into ChatGPT/Claude. Dictate the full context, constraints, and edge cases—Flow cleans filler, adds punctuation, and formats your thoughts into something models can actually use. Better prompts in seconds, without retyping.
*This is sponsored
THROUGH THE VALLEY
Anthropic has raised $30 billion in a new Series G round, lifting its valuation to about $380 billion, up from $183 billion in its last funding round. The raise was led by Singapore’s GIC and Coatue, with support from Founders Fund, D. E. Shaw Ventures, MGX, Accel, General Catalyst, Jane Street, and the Qatar Investment Authority. The funding comes as Anthropic competes more directly with OpenAI for large enterprise customers. OpenAI is also reported to be seeking up to $100 billion in fresh funding. Anthropic says demand for its Claude models is rising fast across major companies.
Spotify says AI has changed how its top engineers work. On its Q4 earnings call, co-CEO Gustav Söderström said some developers have gone months without writing code themselves, relying instead on an internal AI model called Honk, built on Claude Code. Engineers can request bug fixes or new features from Slack and receive updated app builds in near real time, even while commuting. Spotify shipped more than 50 product updates in 2025, including AI features like Prompted Playlists and About This Song. The company also pointed to its music data advantage and said it continues to filter AI-generated tracks used for spam.
Google DeepMind says its Gemini Deep Think model can now handle research-level math and programming tasks, building on its gold medal result at the 2025 International Math Olympiad. The system uses a research agent that works through problems step by step, checks its own answers, fixes errors, and stops when it cannot find a valid solution. New versions show strong gains on PhD-level benchmarks and have already helped produce publishable research in math, computer science, and physics, though no major breakthroughs have been announced. Google claims Gemini 3 Deep Think now outperforms Claude Opus 4.6 and GPT-5.2 and is available in the Gemini app for Ultra users and select partners via API.
OpenAI has launched GPT-5.3 Codex Spark, a fast coding model designed for quick, interactive tasks. It runs on Cerebras chips, making it OpenAI’s first major product not built only on Nvidia hardware. Spark can generate over 1,000 tokens per second, which makes it much faster than earlier Codex models, though it trades some accuracy for speed. OpenAI plans to use Spark for rapid code edits and small changes, while the full Codex model handles longer, background tasks. The model is rolling out to ChatGPT Pro users, with limited access for enterprise API customers.
Interest in ByteDance’s SeeDance 2.0 is reaches all time high. Early tests and viral clips show the model producing videos that look very close to real footage. Videos made using simple prompts have spread widely online, and even experts and creators have said some are hard to distinguish from real shots. The Motion Picture Association has criticized the tool after a hyper-realistic clip showing digital versions of Tom Cruise and Brad Pitt fighting went viral, sparking debate about copyright and how believable AI video has become. These reactions underline how quickly AI video realism is blurring the line between real and fake.
TRENDING TOOLS
Simile > A simulation platform for human behavior
Hibiki-Zero > An open-source model that translates French, Spanish, Portuguese, or German speech to English in real-time
ChatCut > Seedance 2.0, available inside this AI
HappyCappy > The agent-native computer, in your browser, and now on your phone (this is quite similar to Manus AI)
Monaco > It automates customer acquisition and revenue growth for startups
Wabi > The first personal software platform
WHAT I'M CONSUMING
INDUSTRY MOVES
Simile launched with $100M to simulate human societies using AI agents modeled on real behavior. Andrej Karpathy participated as an angel investor, describing it as a new paradigm for using LLMs as population-scale simulators.
Apptronik raises $935M at $5.5B+ valuation to commercialize cheaper humanoid robots
THE VALLEY GEMS
THAT’S ALL FOR TODAY
Thank you for reading today’s edition. That’s all for today’s issue.

💡 Help me get better and suggest new ideas at [email protected] or @heyBarsee
👍️ New reader? Subscribe here
Thanks for being here.
HOW WAS TODAY'S NEWSLETTER |
REACH 100K+ READERS
Acquire new customers and drive revenue by partnering with us
Sponsor AI Valley and reach over 100,000+ entrepreneurs, founders, software engineers, investors, etc.
If you’re interested in sponsoring us, email [email protected] with the subject “AI Valley Ads”.



