| Hard budgets (team / key / user) (scored) | yes AI Rate Limiting Advanced (Enterprise plugin) enforces token or dollar-cost limits per consumer group |
|---|---|
| Rate & spend limits (scored) | yes Token and cost strategies; model input/output cost configured per 1M tokens |
| Cost attribution / chargeback tags (scored) | yes Consumers and consumer groups; Konnect Metering & Billing lists showback and token budgeting |
| FOCUS export (scored) | unknown No FOCUS export found in Kong docs |
| EU data residency option (scored) | yes Dedicated cloud gateways in a region of your choice; self-hosted option Source: konghq.com/pricing |
| Public pricing (scored) | partial Source: konghq.com/pricing |
| Open source | partial Kong Gateway is Apache-2.0 with free AI plugins (e.g. AI Proxy); AI Rate Limiting Advanced and AI Semantic Cache are paid plugins |
| Hosting | Both Konnect serverless, dedicated cloud (AWS/GCP/Azure region of choice) or hybrid; fully self-hosted via Gateway Enterprise Source: konghq.com/pricing |
| Supported providers / models | 16 AI Proxy list: Anthropic, Azure OpenAI, Bedrock, Cerebras, Cohere, DashScope, Databricks, DeepSeek, Gemini, Hugging Face, Llama, Mistral, Ollama, OpenAI, vLLM, xAI |
| Caching | yes AI Semantic Cache (Enterprise plugin): exact and semantic, Redis/Valkey or pgvector |
| Pricing model | Usage-based (Konnect Plus) or custom Enterprise Source: konghq.com/pricing |
| Public price points | Konnect Plus: serverless AI gateway control plane $25/mo; hybrid $200/mo; dedicated $500/mo + $0.15/GB; 1M requests incl., $200 per extra 1M (max 10M/mo); $100/mo per proxied model; AI paid plugins as add-on (price not public) Source: konghq.com/pricing |
Not verified yet: FOCUS export. "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.