Motif / motif/motif-3

Motif 3 - access through LLMTR

Motif 3 is Motif's text model for agentic coding, tool-driven work and terminal-style troubleshooting. It has a 262,144-token context window and there is no separate output limit — the response shares that window, so a long prompt leaves less room for the answer. It accepts text only: image, audio and document input are rejected. It supports tool/function calling (including forced and named tool choice), JSON mode, schema-enforced output, streaming and prompt caching. The model always reasons before answering and this cannot be turned off; those hidden tokens are billed as output. Two parameters to avoid: tool_choice "none" is rejected when tools are attached — use "auto" or "required", or send no tools; and a stop sequence can cut the reply to nothing. The model is currently free, under a request quota counted over a rolling 24-hour window.