4 articles
Browse 4 articles, tool introductions and practical notes about Open-source models, ordered by publication date.
An uncensored DeepSeek V4.1 Flash weight release claims 320/320 HarmBench compliance, but its own MMLU score drops 3.86 points and testing requires four H200 GPUs.
Qwen releases Qwen3.6-35B-A3B with enhanced agentic coding and frontend workflow capabilities. New thinking preservation feature maintains context across conversations. Built on Qwen3.5 architecture with MoE + sparse activation, supports 201 languages, Apache 2.0 licensed.
Baidu ERNIE-Image: 8B parameter open-source text-to-image AI model achieves SOTA performance. GenEval 0.8856, LongTextBench 0.96+. Superior text rendering, runs on 24GB VRAM, 8-step Turbo mode.
Google releases Gemma 4 multimodal models with dense and MoE architectures, 256K token context, native reasoning mode, and deployment from mobile to server. Four model sizes available under Apache 2.0.