Pricing
prxy.monster prices the gateway, not your provider tokens. API-key users bring provider keys; providers bill those users directly. prxy.monster never marks up BYOK provider inference.
Comparison first
Public pricing mechanics as of May 4, 2026:
| Question | prxy.monster | OpenRouter | Portkey | Helicone | LiteLLM |
|---|---|---|---|---|---|
| Who bills provider inference on BYOK? | Provider bills the customer directly. prxy does not settle or mark up BYOK inference. | OpenRouter says provider prices pass through; BYOK has a post-free-tier fee. | Public plans price recorded logs / requests and gateway features. | Public plans price subscription plus usage-based requests/storage. | Self-hosted OSS: operator pays providers directly. |
| Hosted billing unit | Gateway requests. | Model token prices / credits. | Recorded logs / requests. | Requests, storage, and plan tier. | OSS self-host; hosted/enterprise separate. |
| Provider-token markup on prxy-style BYOK | None. No provider-token line item from prxy. | FAQ says no inference markup; BYOK after 1M requests/month has a 5% fee tied to OpenRouter-equivalent model cost. | No public provider-token markup claim found; plan is not priced as prxy-style request-only BYOK. | No public provider-token markup claim found; plan includes usage-based gateway/observability costs. | No hosted markup in OSS mode. |
| Public itemized payment ledger | receipts.prxy.monster for settled agent-commerce receipts. | Not a public ledger. | Not a public ledger. | Not a public ledger. | N/A. |
Sources: OpenRouter FAQ, OpenRouter pricing, Portkey pricing, Helicone pricing, LiteLLM GitHub.
How you pay (pay-per-use first)
Start free, pay only when you grow. No monthly plan required.
- Free tier — 5,000 gateway requests/month, no card.
- Pay as you go — top up credits in Lair when you exceed free; billed per request, not per token.
- Volume packs (optional) — Pro/Team subscriptions for teams that want included request bundles + lower overage rates.
Provider inference is always separate: your Anthropic/OpenAI/Google key, your provider invoice, zero token markup from prxy.
Two billing paths
| Path | Auth | Provider key | Provider invoice | prxy invoice |
|---|---|---|---|---|
| API-key BYOK | Authorization: Bearer prxy_... | You register or supply it | You pay the provider directly | Free allowance, then pay-per-use credits or optional volume pack |
| Managed MPP | Authorization: Payment ... | Not required from agent | Included in MPP price | $0.50 direct settled call, or effective $0.05 per call inside a funded session |
API-key routes do not mark up inference. Managed MPP is different: the agent may have no prxy account and no provider key, so the MPP price includes both PRXY pipeline execution and the upstream model call.
Receipts
For API-key BYOK routes, prxy billing and provider billing are separate: prxy bills requests, providers bill tokens. Lair shows prxy request usage; your provider dashboard remains the source of truth for provider tokens.
For managed MPP and future merchant endpoints, settlement returns an itemized Payment-Receipt header and the receipt can be looked up at receipts.prxy.monster/r/<id>. MPP discovery, the HTTP 402 challenge flow, /v1/agent/quote, /v1/agent/sessions, production settlement, and Payment-Receipt headers are live for the managed MPP route.
Tiers
| Tier | Cost | Included usage | Overage | Production features |
|---|---|---|---|---|
| Free | $0 | 5,000 requests/mo | None; hard cap | Full pipeline trial, local mode, community support |
| Pay as you go | Top-up credits | After free allowance | Per-request debit from balance | No subscription; auto-recharge optional |
| Pro (optional volume) | $20/mo | 100,000 requests/mo | $0.20 / 1k requests | All modules, analytics, MPP merchant preview |
| Team (optional volume) | $99/mo | 1,000,000 requests/mo | $0.10 / 1k requests | SSO, audit, shared policies, priority support |
| Enterprise | Custom | Custom | Custom | Dedicated support, private deployment options, custom contracts |
One request means one HTTP call into the gateway. Streaming counts as one. Cached hits count as one. Failed upstream calls do not count.
Production, preview, planned
| Surface | Status | Notes |
|---|---|---|
API-key BYOK /v1/messages | Production | Anthropic Messages-compatible route. |
API-key BYOK /v1/chat/completions | Production | OpenAI Chat Completions-compatible route. |
| Built-in module pipeline | Production | See module matrix. |
| Local mode | Production | MIT local edition; no billing or cloud sync. |
| MPP discovery + 402 challenge | Live protocol surface | /.well-known/mpp and /v1/agent/messages challenge flow. |
| MPP production settlement | Production | SPT verification, charging, settlement, managed inference, and Payment-Receipt headers are live for the managed MPP route. |
| ATXP MCP, native x402, Tempo | Blocked / planned | Separate payment rails. Do not market them as live until their readiness objects mark production_live. |
| MPP merchant endpoints for customers | Planned | Do not rely on it for production customer endpoints yet. |
| OpenAI Responses / Assistants / Realtime APIs | Planned or unsupported | See compatibility. |
Why no token markup on BYOK?
Token markup compounds with scale and makes cost attribution muddy. BYOK keeps the provider relationship direct: your provider key, your provider limits, your provider invoice.
prxy.monster charges for the gateway work around the provider call: pruning tool bloat, cache lookup, context preservation, pattern injection, routing, and spend controls.
Local mode
Local mode is free forever. It runs the gateway on your hardware with local storage and no hosted billing surface.
Local mode is not ad-supported or telemetry-supported. The container does not phone home. See Local mode privacy.