Is DeepSeek-V3.2 (Non-thinking Mode) or Kimi K2.5 cheaper?

DeepSeek-V3.2 (Non-thinking Mode) is cheaper on output tokens ($0.42 vs $3.02 per 1M).

Which has the larger context window, DeepSeek-V3.2 (Non-thinking Mode) or Kimi K2.5?

Kimi K2.5 has the larger context window (262K tokens).

DeepSeek-V3.2 (Non-thinking Mode) vs Kimi K2.5

DeepSeek-V3.2 (Non-thinking Mode) is cheaper on output tokens, while Kimi K2.5 offers a larger context window. Choose DeepSeek-V3.2 (Non-thinking Mode) or Kimi K2.5 based on the trade-off between cost, context, and the benchmarks that matter for your use case.

Spec	DeepSeek-V3.2 (Non-thinking Mode)	Kimi K2.5
Provider	ZenMux	ZenMux
Input / 1M tokens	$0.28	$0.58
Output / 1M tokens	$0.42	$3.02
Context window	128K	262K
Parameters	—	1T
Open weights	No	No
Released	Dec 2025	Jan 2026

DeepSeek-V3.2 (Non-thinking Mode) details →Kimi K2.5 details →

FAQ

Pricing is indicative — confirm with the provider before production use. Highlighted values indicate the better figure for that row.