Variable RECOMMENDED_MODELSConst

Type Declaration

• Readonlyfast: "gemma-3n-e2b"

  Fast responses, simple tasks (~2GB RAM)

• Readonlybalanced: "gemma-3n-e4b"

  Best quality/speed balance (~3GB RAM)

• Readonlyquality: "gemma-3-27b"

  Maximum quality (~18GB RAM)

• Readonlyedge: "gemma-3n-e2b"

  Best for edge/mobile - minimal RAM

• Readonlymultilingual: "qwen3-8b"

  Best multilingual

• Readonlyreasoning: "deepseek-r1-14b"

  Complex reasoning (chain-of-thought)

• Readonlycode: "qwen-2.5-coder-7b"

  Code generation

• ReadonlylongContext: "gemma-3-27b"

  Long documents (128K context)