Kimi K2.6

Stable3 articles3 sourcesDernière mise à jour: 25/04/2026

Sujets associésmoonshot ai artificial intelligence open source kimi chinese tech deepseek ai models benchmark

Derniers articles

25/04/2026Résumé IA

DeepSeek V4 Pro Ranks Second Behind Moonshot AI's Kimi in Global Open-Source Benchmark

DeepSeek's V4 Pro model ranked second globally among open-source AI models, trailing Moonshot AI's Kimi K2.6, according to benchmark firm Artificial Analysis. While V4 Pro showed improvement over its predecessor V3.2, it still lagged behind top competitors, highlighting challenges facing Chinese AI firms racing to close the gap with the US amid intensifying competition and computing power constraints.

Chinese AI Startup Moonshot AI Releases Open-Source Model Kimi K2.6

Moonshot AI has released Kimi K2.6, its latest open-source flagship model featuring upgrades in long-horizon coding, motion-rich front-end generation and agent-based workflows. The company claims the model performs on par with or better than GPT-5.4, Claude Opus 4.6 and Gemini 3.1 Pro across benchmarks, though independent verification remains limited. Artificial Analysis ranks it among the top tier of open-source models globally.

Kimi K2.6: Açık Kaynaklı Yapay Zeka Modeli 300 Ajanlı Swarm Mimarisiyle Tanıtıldı

Kimi, yeni açık kaynaklı modeli K2.6'yı duyurdu. Model, Rust, Go ve Python dillerinde %12 daha yüksek kod üretim doğruluğu ve %18 uzun bağlam kararlılığı artışı sunuyor. Agent Swarm mimarisi 300 farklı alt ajana kadar ölçeklenebiliyor. 13 saatlik çalışmayla 4.000 satır kod değiştirerek finansal motorun performansını %185 artırdı. Araç çağırma başarı oranı %96,60.

ShiftDelete

2 min de lecture