Fallback Models
Claude Code that never stops, on any model
When Anthropic goes down, your plan limit hits, or new credit policies kick in, Edgee keeps your Claude Code session running by routing to a fallback model automatically. Same prompt, same flow, zero code changes.
Works with Claude Code, Codex, OpenCode. Team plan required for full fallback.
$
Backed by founders and leaders of
Three failure scenarios every Claude Code user has experienced
You cannot ship when your model stops responding. Here is what happens in practice.
Anthropic outage mid-task
You are deep in a refactor and Claude Code stops responding. Status page says degraded. Your flow is broken, deadline looms.
Weekly plan limit reached
You hit your Opus cap on Tuesday. Four days of Sonnet or nothing until the reset. Your sprint plan does not care about the calendar.
Credit policy changes
Anthropic is moving to credit-based billing starting June 15, 2026. This shifts usage limits and introduces new quota mechanics. Teams will need a Plan B.
How it works in your dashboard
Configure once. Failover automatically.
In your Edgee dashboard, you set a priority-ordered model chain. Edgee handles the rest.
Fallback on outage
When the primary model returns a 429 or 5xx, Edgee instantly retries your request through the next configured fallback in the chain. Your Claude Code session keeps running.
Fallback on rate limit
Hit a weekly plan cap or a hard rate limit? Edgee detects the exhausted quota and transparently routes to a fallback model that is available and fast.
Always-on smart routing
Reroute models lets you always send requests to a specific model, regardless of what the client asked for. Use it for cost optimization or standardizing on a single provider fleet-wide.
Claude Code request
alex-chen · implement-feature.ts
Primary: Claude Opus
Weekly limit reached — 0 tokens remaining
Fallback 1: Mistral Large
Routed · 312ms to first token
Coding continues
Transparent to the dev
Supported models
6 Edgee-hosted models, ready to fall back
All available out of the box. No API keys needed. New models added regularly.
Gemma 4 26B
GLM-5
ZAI
Qwen3 Coder 480B
Qwen
Kimi K2.5
Moonshot AI
MiniMax M2.5
MiniMax
Qwen3 Coder Next
Qwen
Plus your BYOK models
OpenAI, Anthropic, Mistral, DeepSeek, xAI, and more
Bring Your Own Cloud
One-click fallback to Bedrock, Vertex, or Azure
Same Claude Code setup. Zero code changes. Your cloud account, your data. You paste credentials once. Edgee routes your fallback traffic through your own cloud provider with the same latency and transparency as our hosted models.
- AWS Bedrock credentials keyed by region
- Google Vertex AI service account JSON
- Azure OpenAI endpoint + API key
- Credentials tested live before going active
AWS Bedrock
Multi-region credentials via access key. Edgee routes to Bedrock models in your AWS account.
Google Vertex AI
Service account JSON from Google Cloud Console. One paste, Edgee resolves OAuth2 tokens at runtime.
Azure OpenAI
Endpoint URL plus API key. Edgee derives the models endpoint from your deployment automatically.
How it works
Three steps. No config files. No proxy setup. No code changes.
- 1
Claude Code calls Anthropic through Edgee
Install the Edgee CLI once. Claude Code sends requests through the Edgee Agent Gateway instead of calling Anthropic directly. No code changes.
- 2
Edgee detects failure or threshold
If Anthropic returns a 429, 5xx, or your plan cap triggers, Edgee flags the request within milliseconds without showing an error to the user.
- 3
Edgee routes to your configured fallback
You set a priority-ordered model chain in the dashboard. Edgee routes the same request to the next available model. Same Claude Code session, zero interruption.
Anthropic returns 429
Rate limit reached
Edgee routes to GLM-5
Fallback model from your chain
Why this matters now
Anthropic credit-based billing shift takes effect June 15, 2026. This introduces credit quotas, new rate-limiting mechanics, and a different relationship between your spend and your access. If you depend on Claude Code daily, you need a Plan B that works without touching your workflow.
Edgee Fallback Models is not anti-Anthropic. It is a rational layer of resilience. We built this because we use Claude Code ourselves and could not afford downtime.
Without fallback vs with Edgee
What a Claude Code outage looks like before and after.
| Claude Code alone | Claude Code + Edgee Fallback | |
|---|---|---|
| Downtime handling | Manual restart | Automatic fallback within ~300ms |
| Rate limit recovery | Wait for reset | Instant failover to next model |
| Model choice | One provider only | 6+ Edgee-hosted models + BYOK |
| Setup time | — | < 2 minutes in dashboard |
| Fallback cost visibility | None | Tracked separately, billed at lower rates |
Pricing
Fallback Models is included in Team
Automatic fallback and rerouting is a Team plan feature. Start with a 14-day free trial. No credit card required.
Team plan
For teams that cannot afford to stop coding when a provider goes down.
per developer / month
Never lose your coding flow again.
Install in 30 seconds. Works with your existing Claude Code setup. No credit card.
Questions devs actually ask
Part of the Edgee Agent Gateway
Routing is one of three pillars.
Edgee also compresses tokens at the edge (up to 50% savings) and observes every session at team level with per-repo attribution.