system-one: batched single-token choice inference

Constrains token selection to choice indices so any open LLM answers many questions in one pass; includes Doom and Wikipedia-racing demos on Qwen models.

Notes

172 ms versus 600 ms median action latency compared with tool calling on Qwen3-8B.

source-backed — Public repository, docs or live artifact. About this label

Destinations

Filed under

• #replica
• #qwen
• #doom
• #github

In collections

• Games & simulations →
• Local & replica models →