vs
One table to see how kimi-k3 (Moonshot) and kimi-k2.7-code (Moonshot AI) differ. On EasyAPI both use the same endpoint and the same key; only the model name changes.
Quick verdict
- kimi-k2.7-code has the lower input price, 72% cheaper than kimi-k3.
- kimi-k2.7-code has the lower output price, 76% cheaper than kimi-k3.
Specs and pricing side by side
| Dimension | kimi-k3 | kimi-k2.7-code |
|---|---|---|
| Provider | Moonshot | Moonshot AI |
| Type | Text | Text |
| Input price | ¥15.8346 | ¥4.503 |
| Output price | ¥79.1728 | ¥18.7046 |
| Cache read | ¥1.5835 | ¥0.9006 |
| Cache write | — | ¥5.2964 |
| Capabilities | 文本 · 多模态 | 文本 · 多模态 |
| Endpoints | 1 | 1 |
Platform prices per million tokens (image / video models use their own units); your invoice is authoritative.
kimi-k3
多模态推理模型，面向复杂编程、知识工作与长程 Agent 流程。擅长在大型代码库里调用工具、调试，并读取图像、日志与测试反馈来修正结果。
kimi-k2.7-code
Kimi K2 系列编程模型，接受文字与图像输入，默认开启思维链，适合长流程 Agent 编程与多轮调试。
FAQ
- Which is cheaper, kimi-k3 or kimi-k2.7-code?
- kimi-k2.7-code has the lower input price, 72% cheaper than kimi-k3.
- Can I call kimi-k3 and kimi-k2.7-code through the same API?
- Yes. Both are served through EasyAPI's OpenAI-compatible API. Point base_url at EasyAPI and swap the model name; no separate vendor accounts needed.