You are viewing a single comment's thread from:

RE: LeoThread 2025-03-06 10:50

in LeoFinance3 days ago

Part 4/10:

Andre Karpathy, a former researcher at OpenAI turned founder of Eureka Labs, had early access to Grock 3 and found it consistently outperformed DeepSeek R1, especially in logistical reasoning tasks. For instance, Grock 3 effectively created a hexagonal grid for the board game Settlers of Catan—an area where other models, including DeepSeek R1 and Gemini 2.0, fell short.

Benchmark Testing: Grock 3 vs. DeepSeek R1