On Monday, Anthropic announced the launch of Claude 3.7 Sonnet, the most intelligent model to date and the first hybrid reasoning model on the market. The company revealed that it tested the latest model on the Game Boy classic Pokémon Red.
The company revealed in a blog post on Monday that the model was equipped with basic memory, screen pixel input, and function calls to press buttons and navigate around the screen.
Anthropic Enters into AI Race With Claude 3.7
What sets Claude 3.7 Sonnet apart is its ability to engage in “extended thinking”; which is a capability linked to OpenAI’s o3-mini and DeepSeek’s R1, where the AI applies more computing power over time to solve complex problems. This feature proved particularly useful in Pokemon Red, where strategic decision-making is crucial.
Amid the growing dominance of new AI models like DeepSeek, the results of Clause 3.7 were incredible. Unlike its predecessor, Claude 3.0 Sonnet, which failed to leave the house in Pallet Town (where the game begins), Claude 3.7 Sonnet successfully defeated three Pokemon gym leaders and earned their badges.
However, the key details about the new AI model remain undisclosed. While Anthropic stated that the model performed 35,000 actions to reach the third gym leader, Lt. Surge, the company has not revealed the computational resources required or the time it took to achieve these milestones.
This unconventional benchmarking method shows the growing capabilities of AI in reasoning and long-term sequential decision-making. As AI models become more sophisticated, it won’t be long before developers push the limits of their gaming prowess even further.
Source: https://www.anthropic.com/news/claude-3-7-sonnet
Latest Stories:
Microsoft’s Data Center Lease Cancellations Raise Concerns Over AI Growth
Wall Street Analysts Warn of Steep Declines for 2 Major AI Stocks
