# Source: https://huggingface.co/Motif-Technologies/Motif-3 (accessed
# 2026-09-14): repo created 2026-08-07, 155 safetensors shards, ungated,
# MIT license.
# tool_call: the model card documents agentic tool use and ships a vLLM
# --tool-call-parser motif; live probe on 2026-09-14 returned a real
# tool_calls response (finish_reason = tool_calls).
# temperature: the card evaluates at sampling temperature 1.0; live probe on
# 2026-09-14 accepted 0.2 and 1.7, with 1.7 visibly changing the output.
name = "Motif 3"
family = "motif"
description = "Motif 3 is a large-scale, decoder-only Mixture-of-Experts (MoE) language model with 314 billion total parameters and 13.2 billion parameters activated per token."
release_date = "2026-08-07"
last_updated = "2026-08-07"
attachment = false
reasoning = true
temperature = true
tool_call = true
structured_output = false
open_weights = true
license = "MIT"

[limit]
context = 262_144
output = 262_144

[modalities]
input = ["text"]
output = ["text"]

[[weights]]
label = "Model weights"
url = "https://huggingface.co/Motif-Technologies/Motif-3"
format = "safetensors"