Kimi K2.6 Beats Claude, GPT-5.5, and Gemini in Coding Contest
Original: Kimi K2.6 just beat Claude, GPT-5.5, and Gemini in a coding challenge
Why This Matters
Demonstrates rising competitiveness of Chinese AI models against Western frontier labs
Chinese AI model Kimi K2.6 from Moonshot AI won an AI coding contest, scoring 22 match points with a 7-1-0 record against major models. The Word Gem Puzzle challenge pitted 9 models against each other in sliding-tile puzzles with 10-second time limits.
In the AI Coding Contest's Day 12 Word Gem Puzzle challenge, Kimi K2.6 from Chinese startup Moonshot AI defeated prominent Western AI models. The open-weights model achieved 22 match points with a 7-1-0 record, followed by Xiaomi's MiMo V2-Pro in second place with 20 points. GPT-5.5 placed third with 16 points, while Claude Opus 4.7 finished fifth with 12 points. The challenge involved sliding-tile letter puzzles on grids ranging from 10x10 to 30x30, where models had to form valid English words within 10-second time limits. Kimi won through aggressive sliding strategies, scoring moves that unlocked positive-value words. The scoring system rewarded longer words while penalizing shorter ones, with words under seven letters costing points.