# Opt in with chat_template_args.enable_thinking=true. Baseten documents no
# explicit false behavior, effort values, or reasoning-token budget.
# https://docs.baseten.co/inference/model-apis/reasoning

name = "GLM 5"
description = "Flagship GLM model for hybrid reasoning, coding, and agentic engineering"
family = "glm"
release_date = "2026-02-12"
last_updated = "2026-02-12"
attachment = false
reasoning = true
temperature = true
tool_call = true
structured_output = true
knowledge = "2026-01"
open_weights = true

[[reasoning_options]]
type = "toggle"

[interleaved]
field = "reasoning_content"

[cost]
input = 0.95
output = 3.15
cache_read = 0.2

[limit]
context = 202_800
output = 202_800

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