Anthropic models fail on OpenCode Zen

// 2h agoINFRASTRUCTURE

Anthropic models fail on OpenCode Zen

Jamie Turner (co-founder of Convex) reported experiencing persistent issues over several days when trying to access Anthropic models via OpenCode Zen (OC Zen), a curated model gateway for coding agents. While Anthropic models were failing to respond or function through the gateway, OpenAI's GPT models continued to work properly, pointing to a service disruption or routing issue specific to Anthropic integration on OpenCode's infrastructure.

// ANALYSIS

API model gateways are only as strong as their weakest integration, and when premium coding models like Claude go down on curated gateways, developers will quickly revert to direct API integrations or alternatives like GPT.

* High-reasoning coding agents are heavily reliant on Anthropic's Claude models, making any gateway downtime for these specific models highly disruptive to developer workflows.

* The fact that GPT models remained functional suggests the issue is isolated to OpenCode's Anthropic routing/API key management or Anthropic's specific API endpoints rather than a total gateway outage.

* This highlights the vulnerability of relying on multi-model aggregators/gateways for mission-critical developer environments over direct-to-provider APIs.

// TAGS

opencodeanthropicclaudeapi-gatewaydeveloper-toolsartificial-intelligence

DISCOVERED

2h ago

2026-06-21

PUBLISHED

2h ago

2026-06-21

RELEVANCE

6/ 10

AUTHOR

jamwt

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

UPDATE56m ago

Executor v1.5.16 lands Microsoft Graph integration

Executor v1.5.16 introduces Microsoft Graph integration with support for OAuth and multiple accounts. The update also improves attachment handling, allowing agents to invoke emit() to render files and images directly in the chat interface.

NEWS1h ago

Anthropic develops restricted Mythos successor

Anthropic is reportedly developing a next-generation AI model designed to surpass the capabilities of its restricted Mythos architecture. The new model is expected to serve as a successor to the highly advanced baseline, which was previously kept from public release due to safety and cybersecurity concerns.

UPDATE3h ago

Nuke on Rails expands security auditing skill

Nuke on Rails has expanded its AI coding agent security auditing skill from 9 to 15 checks to cover the OWASP Top 10. The update adds checks for WebSocket access flaws, exposed dashboards, data-leaking webhooks, and AI-specific risks like prompt injection.