Kimi K2.7 Code vs Kimi K3 1M

Moonshot's Kimi K2.7 Code against Moonshot's Kimi K3 1M — price, context and what each is best at. Both are available through one OpenAI-compatible API.

AnyModel — you pay	Kimi K2.7 Code	Kimi K3 1M
Vendor	Moonshot	Moonshot
Input / 1M	$0.075	$0.05
Output / 1M	$0.075	$0.05
Context	256K	1M
Best for	Coding agents	Long documents

Kimi K2.7 Code

Moonshot's strongest coder, always-thinking.

Kimi K2.7 Code is Moonshot's coding specialist, served through the connected Kimi Code subscription. It is an always-thinking model tuned for agentic software engineering with a 256K window.

More about Kimi K2.7 Code →

Kimi K3 1M

Moonshot's flagship with the full million-token window.

Kimi K3 1M is Moonshot's newest flagship — coding, image understanding and long-horizon reasoning — served over our own metered capacity with the model's full 1M-token context instead of the 256K the Kimi Code subscription route allows.

More about Kimi K3 1M →

Quick take: pick Kimi K3 1M when cost matters most; pick Kimi K3 1M when you need the largest context. With AnyModel you don't have to choose up front — switch the model id any time.

FAQ

Which is cheaper, Kimi K2.7 Code or Kimi K3 1M?

Kimi K3 1M is cheaper on output tokens ($0.05 vs $0.075 per 1M).

Which has a larger context window, Kimi K2.7 Code or Kimi K3 1M?

Kimi K3 1M has the larger context window (1M).