Repository: localaiLicense: apache-2.0
AREX-Turbo is BAAI's compact 4B deep-research agent, fine-tuned from Qwen3.5-4B for long-horizon search, evidence aggregation, constraint verification, and tool-assisted reasoning. It supports text and image input with a 262K-token context window. This entry uses the recommended Q4_K_M GGUF quantization.
Links
Tags
Repository: localaiLicense: apache-2.0
AREX-Turbo is BAAI's compact 4B deep-research agent, fine-tuned from Qwen3.5-4B for long-horizon search, evidence aggregation, constraint verification, and tool-assisted reasoning. It supports text and image input with a 262K-token context window. This entry uses the higher-quality Q8_0 GGUF quantization.
Links
Tags
Repository: localaiLicense: apache-2.0

Jan-Nano is a compact 4-billion parameter language model specifically designed and trained for deep research tasks. This model has been optimized to work seamlessly with Model Context Protocol (MCP) servers, enabling efficient integration with various research tools and data sources.
Links
Tags
OpenResearcher-30B-A3B is an agentic deep-research model with 30B total parameters and 3B active parameters. This entry uses the Q4_K_M GGUF.
Links
Tags
OpenResearcher-30B-A3B is an agentic deep-research model with 30B total parameters and 3B active parameters. This entry uses the Q8_0 GGUF.
Links
Tags