SYLIN

Groot

by Groot Team · Model · Language & code models

Four pruned and fine-tuned MoE models: 19B (3B active) from Qwen 3.6 35B and 280B from DeepSeek 4.1 at half its size; Q4 GGUF of 19B is 12 GB

AccessOpen weightsNeeds~12 GBSize280B parametersLicenseApache 2.0

Evidence

  1. Nov 1
    Groot 19B claimed to beat Qwen 3.6 35B on agentic coding, automation and knowledge benchmarks at half the size

Cost to run

Alternatives

Made from