Palo Alto Networks closed its acquisition of Portkey on 29 May 2026; product is now Prisma AIRS AI Gateway (GA 16 Jul 2026). Source
| Hard budgets (team / key / user) (scored) | yes Usage-limit policies on cost, tokens or requests, grouped by metadata (e.g. per user); requests blocked with HTTP 412. Listed as Enterprise on the pricing page |
|---|---|
| Rate & spend limits (scored) | yes "Granular Budget & Rate Limits" in Enterprise Source: portkey.ai/pricing |
| Cost attribution / chargeback tags (scored) | yes Metadata keys attribute cost to teams, users, features |
| FOCUS export (scored) | no No FOCUS export documented; JSONL log export (Enterprise only, 50k logs per job) |
| EU data residency option (scored) | partial Portkey Enterprise docs describe EU region-pinned SaaS and hybrid; Prisma AIRS AI Gateway deploy docs currently name the Americas region |
| Public pricing (scored) | partial Source: portkey.ai/pricing |
| Open source | partial Core gateway is MIT-licensed; governance, budgets and log export are in the commercial product Source: github.com/Portkey-AI/gateway |
| Hosting | Both SaaS or hybrid (data plane in your Kubernetes); OSS gateway can be self-hosted |
| Supported providers / models | 1,600+ LLMs (OSS) / 3,000+ LLMs (Prisma AIRS) Counts are models, not providers |
| Caching | yes Simple and semantic caching (Production plan) Source: portkey.ai/pricing |
| Pricing model | Freemium SaaS + Enterprise Source: portkey.ai/pricing |
| Public price points | Developer free (10k logs/mo); Production $49/mo incl. 100k logs, +$9 per extra 100k up to 3M; Enterprise custom Source: portkey.ai/pricing |
"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.