Using an open model feels surprisingly good
2 days ago
- The user has been using Claude and ChatGPT for two years but was never a strong advocate for open software.
- Using opencode on their own inference endpoint (Modal with Kimi K3) felt surprisingly good and freeing, as they owned the endpoint and data remained private.
- The motivation was to avoid upgrading their Claude/ChatGPT plan; instead, they tried Kimi K3, recently launched on Modal managed endpoints.
- Setup took about 5 minutes, and the experience felt like cutting ties to other providers, allowing them to breathe freely.