#kimi k2 context miscalculation?

4 messages · Page 1 of 1 (latest)

versed inlet
#

As the graphic shows, KC thinks Kimi only has 32.8K context window, while the model descriptions says it has 128K.
This resulted in an attempted context condensing which failed.

languid ember
#

Is this using openrouter provider?
If yes please setup a desired provider on advanced settings, otherwise you will get different providers with different context window

versed inlet
languid ember
#

Yeah, but openrouter have different providers maybe one of them has a 33k context window

As you can see here chutes have 33k context window
https://openrouter.ai/moonshotai/kimi-k2

Kimi K2 Instruct is a large-scale Mixture-of-Experts (MoE) language model developed by Moonshot AI, featuring 1 trillion total parameters with 32 billion active per forward pass. It is optimized for agentic capabilities, including advanced tool use, reasoning, and code synthesis. Run Kimi K2 with API