[BUG] API timeout errors from Costa Rica (Latin America) - Works fine from Canada

[murias002]

Describe the bug

Claude Code constantly shows "API Error (Request timed out...)" when running from Costa Rica (Latin America). The same configuration works perfectly from a VPS in Canada (Quebec) with ~0.09s response times.

To Reproduce

1. Run Claude Code from Costa Rica (Central America)
2. Execute any command that requires API communication
3. Observe timeout errors and retry loops

Expected behavior

API should respond within reasonable timeouts regardless of geographic location.

Environment

• Claude Code version: 2.1.6
• OS (Costa Rica): Ubuntu 24.04 / macOS (local machines)
• OS (Canada - working): Ubuntu 24.04.3 LTS
• Location affected: Costa Rica, Central America
• Location working: Quebec, Canada
• ISP (Costa Rica): ICE (Instituto Costarricense de Electricidad)

Error messages

API Error (Request timed out...) - Retrying in X seconds.. (attempt X/10)

Additional context

• Issue started: January 12, 2026
• This coincides with the "Increased rate of errors for Opus 4.5" incident reported on status.claude.com
• The API endpoint (api.anthropic.com / 160.79.104.10) responds instantly from Canada
• Other users from Latin America may be experiencing similar issues
• Workaround: Using a VPS/VPN in North America works, but adds friction to the workflow

Possible causes

1. Regional routing issues to Cloudflare/Anthropic infrastructure
2. Timeout thresholds too aggressive for higher-latency regions
3. Related to the Opus 4.5 incidents on Jan 12-13

Request

Could Anthropic investigate:

1. If there are regional blocks or routing issues affecting Latin America
2. If timeout thresholds can be adjusted for higher-latency regions
3. If the recent Opus 4.5 incidents affected specific regions more than others