Models

Download open weights models fine-tuned specifically for memory.

driaforall/mem-agent

tiles run mem-agent

Updated September, 2025

Based on Qwen3-4B-Thinking-2507, this model was trained using GSPO from Dria, an inference lab dedicated to accessibility, coding, and agentic behavior.

tools

thinking

tilesprivacy/ministral-3

Coming soon

tiles run ministral-3

Download Modelfile →

Releasing January, 2026

First in-house memory model based on Ministral-3B-Instruct from Tiles Privacy. This model was trained with GRPO using Unsloth.

tools

thinking