| Hard budgets (team / key / user) (scored) | partial Usage-based rate limits on token costs per user/tenant header act as token budgets; no currency budgets |
|---|---|
| Rate & spend limits (scored) | yes llmRequestCosts (input/output/total/cached/CEL) + global rate limit |
| Cost attribution / chargeback tags (scored) | partial Token usage keyed by headers/metadata; no chargeback reporting |
| FOCUS export (scored) | unknown No FOCUS export found |
| EU data residency option (scored) | yes Self-hosted Source: github.com/envoyproxy/ai-gateway |
| Public pricing (scored) | yes Source: github.com/envoyproxy/ai-gateway |
| Open source | yes Apache-2.0 Source: github.com/envoyproxy/ai-gateway |
| Hosting | Self-host Kubernetes |
| Supported providers / models | unknown README names OpenAI, Azure OpenAI, Gemini, Vertex AI, Bedrock, Mistral and others; no count |
| Caching | unknown |
| Pricing model | Open source Source: github.com/envoyproxy/ai-gateway |
| Public price points | Free Source: github.com/envoyproxy/ai-gateway |
Not verified yet: FOCUS export, Supported providers / models, Caching. "unknown" means we could not confirm it in public sources. Vendors can send evidence to update a cell; see get verified. Verification never changes the scoring rules.