Anthropic’s AI model, Claude 3.7 Sonnet, is making waves in the gaming and AI communities as it attempts to play through the classic game Pokémon Red on a live Twitch stream. The progress of AI, while slow, has surpassed that of its predecessors, showcasing advancements in AI technology. Claude navigates the game by analyzing screenshots, using pathfinding tools, and maintaining dynamic notes about game mechanics and Pokémon identities.
A few researchers at Anthropic have, over the past year, had a part-time obsession with a peculiar problem.
Can Claude play Pokémon?
A thread: pic.twitter.com/K8SkNXCxYJ
— Anthropic (@AnthropicAI) February 25, 2025
It also has access to parts of the game’s memory to check the status of its party.
Despite these capabilities, the AI encounters difficulties and makes amusing mistakes, such as repeatedly walking into a fence near Brock’s gym in Pewter City. Earlier versions of Claude struggled with the game, with Claude 3.5 Sonnet failing to make significant progress and Claude 3.6 Sonnet only managing to defeat a rival and move beyond Pallet Town.
However, Claude 3.7 Sonnet has already defeated three gym leaders, demonstrating its improved abilities.
Introducing Claude 3.7 Sonnet: our most intelligent model to date. It's a hybrid reasoning model, producing near-instant responses or extended, step-by-step thinking.
One model, two ways to think.
We’re also releasing an agentic coding tool: Claude Code. pic.twitter.com/jt7qQmFWuC
— Anthropic (@AnthropicAI) February 24, 2025
The Twitch stream showcases the AI’s gameplay and thought process, allowing viewers to observe its reasoning capabilities in real-time.
Claude 3.7 Sonnet is now available with Perplexity Pro.
We've tested the model internally for some time now and have observed a noticeable improvement in agentic workflows and code generation.
Try it now by switching your "AI Model" in settings. pic.twitter.com/GZkZqAi4q4
— Perplexity (@perplexity_ai) February 25, 2025
Claude’s progress in Pokémon Red
The Twitch stream showcases the AI’s gameplay and thought process, allowing viewers to observe its reasoning capabilities in real-time.
This has sparked interest among both gaming enthusiasts and AI researchers. Pokémon Red, a role-playing game developed by Game Freak and published by Nintendo in 1996 (Japan) and 1998 (North America and Europe), serves as a benchmark for testing Claude’s abilities. The game presents various puzzles and challenges the AI must work through, providing valuable insights into its problem-solving skills.
As the stream continues, it remains to be seen whether Claude 3.7 Sonnet will complete the game faster than the collective effort of “Twitch Plays Pokémon,” a social experiment from over a decade ago in which millions of people collaboratively controlled the game via Twitch chat. That endeavor took 16 days to finish. Claude’s attempt to master Pokémon Red highlights the growing role of AI in gaming and entertainment while also prompting reflections on how technology shapes our experiences in the digital age.
As AI advances, it will be interesting to see how it tackles increasingly complex challenges and influences the future of gaming and beyond.
Image Credits: Photo by Thimo Pedersen on Unsplash
Johannah Lopez is a versatile professional who seamlessly navigates two worlds. By day, she excels as a SaaS freelance writer, crafting informative and persuasive content for tech companies. By night, she showcases her vibrant personality and customer service skills as a part-time bartender. Johannah's ability to blend her writing expertise with her social finesse makes her a well-rounded and engaging storyteller in any setting.
