Latest Claude 3.7 Sonnet model storms the Pokémon world

Post Thumbnail

Anthropic, one of the leaders in artificial intelligence, presented an unusual approach to testing its latest Claude 3.7 Sonnet model, using the iconic Game Boy game Pokémon Red.

According to information published in the company’s official blog on February 24, researchers equipped the model with basic memory, pixel input processing capability from the screen, and functional calls for button pressing and navigation. This allowed the AI to continuously play Pokémon without additional assistance.

A key advantage of Claude 3.7 Sonnet is the “extended thinking” feature, similar to the capabilities of OpenAI o3-mini and DeepSeek R1. This technology allows the model to “reason” when solving complex tasks, applying additional computational resources and spending more time on analysis.

The results of the experiment were impressive. While the previous version of the model, Claude 3.0 Sonnet, couldn’t even leave the starting house in Pallet Town where the game begins, Claude 3.7 Sonnet successfully battled three gym leaders and received their badges.

To achieve these results, the AI performed 35,000 game actions to reach the last gym leader Lieutenant Surge. However, the company did not disclose exact data on the computing power and time spent completing the game.

Although Pokémon Red may be considered more of an entertainment benchmark, using games for AI testing has a long tradition in the research community. In recent months, a number of new applications and platforms have emerged to test AI models’ gaming abilities on various games – from Street Fighter to Pictionary.

This experiment demonstrates the growing ability of artificial intelligence models to navigate complex interactive environments, understand rules, and strategically plan actions to achieve long-term goals – skills that have broad practical applications beyond the gaming industry.

AIvengo > Reviews > Latest Claude 3.7 Sonnet model storms the Pokémon world
Почитать из последнего
UBTech will send Walker S2 robots to serve on China's border for $37 million
Chinese company UBTech won a contract for $37 million. And will send humanoid robots Walker S2 to serve on China's border with Vietnam. South China Morning Post reports that the robots will interact with tourists and staff, perform logistics operations, inspect cargo and patrol the area. And characteristically — they can independently change their battery.
Anthropic accidentally revealed an internal document about Claude's "soul"
Anthropic accidentally revealed the "soul" of artificial intelligence to a user. And this is not a metaphor. This is a quite specific internal document.
Jensen Huang ordered Nvidia employees to use AI everywhere
Jensen Huang announced total mobilization under the banner of artificial intelligence inside Nvidia. And this is no longer a recommendation. This is a requirement.
AI chatbots generate content that exacerbates eating disorders
A joint study by Stanford University and the Center for Democracy and Technology showed a disturbing picture. Chatbots with artificial intelligence pose a serious risk to people with eating disorders. Scientists warn that neural networks hand out harmful advice about diets. They suggest ways to hide the disorder and generate "inspiring weight loss content" that worsens the problem.
OpenAGI released the Lux model that overtakes Google and OpenAI
Startup OpenAGI released the Lux model for computer control and claims this is a breakthrough. According to benchmarks, the model overtakes analogues from Google, OpenAI and Anthropic by a whole generation. Moreover, it works faster. About 1 second per step instead of 3 seconds for competitors. And 10 times cheaper in cost per processing 1 token.