Minico M1 1105 MLX
8-bit MLX build of Mincofficial/Minico-M1-1105 for Apple silicon (near-lossless at this size).
For full precision, mlx_lm loads the main repo directly: --model Mincofficial/Minico-M1-1105.
pip install mlx-lm
mlx_lm.generate --model Mincofficial/Minico-M1-1105-MLX --prompt "What is 5 + 5 x 5?" --chat-template-config '{"reasoning_effort":"low"}'
Effort levels
| Effort | What it does | Max new tokens we use |
|---|---|---|
off |
No thinking. Answers right away (an empty think block is prefilled). | 1024 |
low |
A short thought, then the answer. | 640 |
medium |
A normal thought. | 1152 |
high (default) |
Longer, more careful reasoning. | 2368 |
max |
The longest reasoning. | 4608 |
The chat template also accepts aliases (none, minimal, xhigh, and others) and enable_thinking=False, which means off.
Without a reasoning_effort, and for unknown values, it uses high (the default since 2026-11-08; it was medium). Recommended sampling: temperature 0.1–0.3, top_k 50, repetition penalty 1.05.
Pick one effort per conversation. Switching the effort, or turning thinking on or off, partway through a chat is not recommended: the model was trained on chats that keep one setting from start to finish. Start a new chat to change it.
License
Minico M1 1105 is a modified version of LiquidAI/LFM2.5-350M and is released under the
LFM Open License v1.0 (see LICENSE). The weights were changed by fine-tuning (SFT, DPO, GRPO) and weight merging, as described
above. Commercial use is allowed for organisations under the revenue threshold set in the license; read it before using the model
commercially.
Built on LFM2.5-350M by Liquid AI. Attribution notices from the base model are kept in LICENSE.