October 18, 2026 DeepSeek's New 763B Model Reads 1M Tokens: Pushing KV Cache Compression to the Edge DeepSeekOpen Source LLMKV CacheLong Context
October 17, 2026 Zero Algorithm Changes: DeepSeek Jumps 11 Points on Terminal-Bench DeepSeekOpen Source LLMAI AgentsKV Cache
August 18, 2026 18GB Local AI on Consumer PC Outperforms Cloud Giants: Alibaba Open-Sources Qwen3.8-27B Open-Source LLMQwen3.8Artificial IntelligenceEdge Computing
