Claude Code vs Gemini CLI with Kimi K2

...... ......

wwgoubuli

12h ago

目前组合

claude code 100 美刀版 opencode + gemini pro 2.5

和很多人的感受可能不一样的是,我认为这两者能力其实在同一层级,但要看是什么样的人使用 以及使用方法。

对于新手来说,或者追求尽可能的自动化的人群,前者的大力出奇迹的方式还是可以的,比后者的自我纠错能力强一点。

对于追求高级copilot模式的用户来说,如果你明确的知道自己要做的事的步骤只是用AI来偷懒,那么后者能力完全不输前者。且吞吐速度和API的稳定性比前者还好一点。

这并不是要比较出一个高下, 我都在用。

virushuo

11h ago

我觉得 aider 还是好过claude code , 因为它不试图节约context ...... 配上 gemini 2.5 pro 的大 context 实在是强

Jul 16, 2025 · 3:32 PM UTC

======

木遥

昨天 14:08 来自 weibo

如果一个古人(aka 任何疫情前的人)听到今天程序员的日常对话:

不知道能听懂百分之几……

======

https://moonshotai.github.io/Kimi-K2/

While Kimi K2 serves as a strong foundation for open agentic intelligence, a general agent uses more advanced capabilities such as thinking and visual understanding. We plan to add these to Kimi K2 in the future.

Limitations

In our internal tests, we've identified some limitations in current Kimi K2 models. When dealing with hard reasoning tasks or unclear tool definition, the model may generate excessive tokens, sometimes leading to truncated outputs or incomplete tool calls.

Additionally, performance may decline on certain tasks if tool use is enabled.

When building complete software projects, one-shot prompting yields performance degradation compared to using K2 under an agentic framework. We are working to address these issues in future releases and looking forward to more feedbacks.