Failover Provider

The Failover provider wraps multiple providers and automatically falls back to the next one in the chain if a request fails. Use it for high-reliability setups where uptime is critical.

Configure VibeCody

Config file (~/.vibecli/config.toml)

[failover]
chain = ["claude", "openai", "ollama"]

When a request fails (timeout, rate limit, server error), VibeCody automatically retries with the next provider in the chain. Each provider in the chain must be configured separately.

Full example

# Primary provider
[claude]
enabled = true
model = "claude-sonnet-4-6"

# Secondary provider
[openai]
enabled = true
model = "gpt-4o"

# Local fallback (always available)
[ollama]
enabled = true
model = "qwen3-coder"

# Failover chain
[failover]
chain = ["claude", "openai", "ollama"]

Use from the CLI:

vibecli --provider failover

How It Works

VibeCody sends the request to the first provider in the chain
If it fails (network error, 429, 500, timeout), the next provider is tried
This continues until a provider succeeds or the chain is exhausted
If all providers fail, the error from the last provider is returned

Best For

Production reliability – ensure AI is always available even if one provider has an outage
Rate limit mitigation – overflow to another provider when rate-limited
Cost optimization – try a cheaper provider first, fall back to a more expensive one
Air-gapped fallback – chain a cloud provider with a local Ollama instance

Troubleshooting

All providers failed

Check that each provider in the chain is properly configured with valid credentials
Run vibecli --provider <name> -c "test" for each provider individually to identify which one works

Provider: Failover