ClawQuest’s AIP2-1.0 Beats Claude in Game-Creation Benchmark

ClawQuest has launched AIP2-1.0, an AI agent designed to create and refine video-game code from instructions sent through Telegram. The product combines conversational development tools with a Command-to-Earn incentive system, allowing players to direct the production process and receive rewards for participating. The approach aims to lower the technical barrier to game creation while turning prompts, revisions and other development activity into an interactive, incentive-based workflow.

In its first disclosed benchmark, AIP2-1.0 posted a higher overall score than Claude Opus 5 while incurring model costs equal to 11.2% of its rival’s, a figure rounded to 12% in the related report’s headline. ClawQuest is positioning the result as evidence of a significant price-performance advantage. However, the available report does not specify the full test suite, sample size or independent validation, leaving broader real-world performance to be established.