Pricing
Simple pricing. The whole AI gateway, included.
Free to try. Pay only for what you use, at a flat 4% on top of your own provider's cost — routing, guardrails, RAG, semantic caching, the MCP gateway and cost analytics are part of the product, not add-ons.
Free
No provider key, no card
£0
- Curated free models — Gemini Flash, Groq-hosted Llama and Mistral Small
- Up to 20 requests per organisation per day
- Explore the full dashboard: routing, AI plugins, MCP directory
- Analytics, team roles and 2FA
Pay as you go
RecommendedBring your own provider keys
+4%
on your provider's own cost
- Every provider Airelay supports (45), any model
- All routing strategies, circuit breaker and feedback API
- All AI guardrails & plugins, org-wide
- MCP gateway and A2A proxy
- Your provider bills you directly — we invoice only the markup
- Direct Debit billing; free models stay free
Enterprise
AI at scale
Talk to us
Pricing agreed with you
- Everything in Pay as you go
- High-volume AI traffic
- Commercial terms negotiated for your usage
- Talk through your architecture before you commit
How the 4% works
A model that costs $10 per 1M tokens from its provider costs $10.40 in total through Airelay — $10 billed by your provider to your own account, and the 4% billed by us. The same flat rate whichever provider or model you route to, visible in your dashboard as it accrues. Semantic cache hits aren't sent to a provider, so they add no provider cost.
Compare plans
| Feature | Free | Pay as you go | Enterprise |
|---|---|---|---|
| AI Gateway | |||
| Curated free models, no provider key | Up to 20 requests/day | Included | Included |
| Bring your own keys for 45 providers | Not included | Included | Included |
| Routing policies (fallback, load balancing, cost, latency, semantic) | Included | Included | Included |
| Circuit breaker & response feedback API | Included | Included | Included |
| AI Guardrails & Plugins | |||
| PII redaction (incl. UK identifiers), prompt & response guards | Included | Included | Included |
| Semantic cache, RAG injection, prompt templates, prompt compression | Included | Included | Included |
| Third-party guardrails (your own vendor account) | Included | Included | Included |
| Token & cost budgets, LLM-as-judge scoring | Included | Included | Included |
| Agents & MCP | |||
| MCP directory, MCP gateway and A2A proxy | Included | Included | Included |
| Platform | |||
| Cost & usage analytics, request log, audit log | Included | Included | Included |
| Vault-backed secrets for AI plugin configuration | Included | Included | Included |
| High-volume AI traffic | Not included | Not included | Talk to us |
| Team roles & invitations, 2FA | Included | Included | Included |
| Billing | No card | Direct Debit | Agreed with you |
Third-party guardrail vendors, and models you route to with your own keys, are billed by those vendors and providers under your own accounts.
Start free, scale when you're ready
Free to sign up, no card needed to try the free models. Connect your own provider keys whenever you're ready.