🚫 Anthropic blocked opencode from the Claude Code API. Was it a good move?
George Hotz (geohot) — the hacker behind tinygrad and comma.ai — published a sharp critique of Anthropic after they blocked opencode, an alternative Claude Code client.
The error opencode users receive:
“This credential is only authorized for use with Claude Code and cannot be used for other API requests.”
📌 What’s at stake?
Anthropic has the right to protect their platform. But Hotz’s argument is strategic: blocking users who are already actively using Claude doesn’t bring them back to Claude Code — it pushes them to other providers.
“You will not convert people back to Claude Code, you will convert people to other model providers.”
⚖️ The AI platform dilemma:
| Strategy | Advantage | Risk |
|---|---|---|
| Open API | Broader ecosystem, more adoption | Loss of UX control and monetization |
| App-locked API | Full product control | Alienating power users and developers |
🔍 Explanation in a nutshell
An API is like a door that lets external apps talk to a service. Anthropic has an AI model (Claude) accessible via API. Some developers built alternative apps (like opencode) that use that same API. Anthropic decided to block those apps, forcing users to use only their official app. The debate is whether that helps or hurts product adoption.
More information at the link 👇
