name = "Kimi K2.5"
description = "Kimi multimodal agent model for visual understanding, coding, and planning"
family = "kimi-k2"
release_date = "2026-01-27"
last_updated = "2026-01-27"
attachment = true
reasoning = true
reasoning_options = [{ type = "toggle" }]
temperature = true
tool_call = true
knowledge = "2024-10"
open_weights = true

[interleaved]
field = "reasoning_content"

[cost]
input = 0.6
output = 3.0
cache_read = 0.08

[limit]
context = 262_144
output = 65_536

[modalities]
input = ["text", "image", "video"]
output = ["text"]