Pay for the platform. Share only when you save.
From $18/mo · No share until you've saved $15 · Max ~$36/mo total on Pro. Free = strict fair-use rate limits; bursts may soft-fail with HTTP 429 — upgrade for higher limits.
Free
See your savings — $0 bill
2M tokens / month
Full MCP tools · strict fair-use rate limits
30 MCP req/min per key · team ceiling 30
Playground: low RPM + daily cap (not multi-user load)
Soft-fail HTTP 429 under burst — upgrade for higher limits
7-day activity logs
Pro
Try Pro — $5 first month
cap ~$36/mo
From $18/mo platform · share capped at ~$36/mo
20M tokens / month
300 MCP req/min per API key
L2 hard cap (β) available
Gateway defaults editable
30-day activity logs
Teams
cap ~$100/mo
From $55/mo platform · share capped at ~$100/mo
100M token pool / month
600 MCP req/min per API key
Higher plan RPM limits for agent workflows
Member tool overrides
Team governance controls
90-day activity logs
Enterprise
200M+ tokens (contract)
1200 MCP req/min per API key
Integrator HMAC budget
Extended log retention
Priority support
Free / Playground uses strict fair-use rate limits. Bursts soft-fail with HTTP 429. Playground is for individual evaluation, not multi-user load; upgrade for higher limits. Teams include higher shared-workspace limits.
Email support@smartgate.network for enterprise or billing questions.
Compare plans
| Feature | Free | Pro | Teams | Enterprise |
|---|---|---|---|---|
| Monthly token cap (L1) | 2M | 20M | 100M | 200M+ |
| Full MCP tools | ||||
| Activity log retention | 7d | 30d | 90d | 180d |
| L2 hard cap (β) | — | |||
| Member tool overrides | — | — | ||
| Team gateway policy | — | |||
| Integrator HMAC budget | — | — | — | |
| Max API keys | 2 | 10 | 30 | 9999 |
| MCP req/min (per API key) | 30 | 300 | 600 | 1200 |
| REST write req/min (team) | 20 | 120 | 300 | 600 |
| MCP team ceiling (req/min) | 30 | 600 | 3000 | 9999 |
| Playground req/min (team) | 10 | 60 | 120 | 300 |
| Fair-use and team limits | Strict fair-use; soft-fail 429 under burst | Higher plan RPM limits | Higher team ceiling / shared-workspace limits | Custom limits by contract |
Pricing from billing-savings-share (B4 SSOT) · caps from plan-catalog · Free rate limits are strict fair-use (HTTP 429 soft-fail under burst; upgrade for higher limits)
How SmartGate pricing maps to the intelligence layer
SmartGate is the intelligence layer between AI and the world: an MCP-native algorithmic gateway, not an LLM proxy, chatbot, or model router. You keep your host model and connect SmartGate as a tool server. Pricing therefore meters platform access and gateway tool traffic—Research, Context, Memory, Control, and Pipeline—rather than charging as if we hosted the model that reasons and chats.
Plans share one story. Free and Playground exist for individual evaluation with strict fair-use rate limits; bursts soft-fail with HTTP 429, limit_scope, and Retry-After when available. Pro adds a higher monthly token pool, optional L2 hard cap, and a platform fee plus savings share model capped around the published Pro total. Teams raise shared-workspace limits for builders who share one gateway. Enterprise is contract-based with a larger baseline token pool—contact sales rather than reading a public calculator as a quote.
Gainshare is deliberate asymmetry: you pay a platform fee first, and a variable share only after a free savings allowance, using the same internal gateway benchmark the product uses for Est. USD saved. Search-class estimated savings count at a reduced weight toward share; fetch, compress, dedup, pipe, and budget-block savings count fully. Team memory writes and Playground-only estimates stay out of the share tax base. That keeps the bill tied to governed tool traffic, not to a restated vendor LLM invoice.
What counts toward the monthly token pool is tool token_used recorded through budget guard and audit—including compress, fetch, search, and other MCP modules. MCP requests-per-minute per API key is separate from Playground RPM and from the shared smart_search/smart_fetch per-team gateway cap documented under tool limits. Hitting a limit is a soft-fail entitlement signal, not a silent drop: reduce request rate or upgrade for higher limits.
Choose Free to evaluate the seven algorithm primitives on one Streamable HTTP endpoint. Choose Pro when a single team needs Usage Reports, stronger caps, and Gainshare with a published ceiling. Choose Teams when multiple builders share workspace limits and audit. Choose Enterprise when retention, pool size, and contract terms need a sales conversation. In every case SmartGate remains five intelligence capabilities powered by seven primitives—not another AI assistant and not a ChatGPT alternative.
Migration and FinOps notes
Moving from ad-hoc fetch scripts or per-agent reimplementations usually means pointing MCP clients at one origin, minting team API keys, and turning on Activity Logs before you debate compression ratios. FinOps reviewers should treat Est. USD saved as a gateway counterfactual for comparing tool paths, then reconcile separately against the host model invoice. Security reviewers should start from audit retention by plan and team_id isolation, then sample traces for a pipe that includes search, fetch, dedup, and context_gate.
Pricing pages intentionally avoid invented customer logos, concurrent capacity figures, and fixed savings percentages. Savings depend on input structure and the ratios you set; the dashboard shows per-request metrics from audit. If you need a contractual commitment beyond published Pro and Teams ceilings, use contact sales for Enterprise rather than extrapolating public tiers.
Pay for the platform. Share only when you save.
From $18/mo · No share until you've saved $15 · Max ~$36/mo total on Pro. Free = strict fair-use rate limits; bursts may soft-fail with HTTP 429 — upgrade for higher limits.
Does SmartGate replace my LLM?
No. SmartGate is the AI intelligence layer — not a chatbot, ChatGPT alternative, or LLM proxy. Research, context, memory, control, and pipeline run as tool traffic; your host model still reasons and chats.
How is my bill calculated?
Platform fee plus a savings share only after your free allowance. Pro bills are capped at about $36/mo total; Teams at about $100/mo.
What is Est. USD saved?
Six gateway tools at an internal $2.50/M input benchmark — a gateway counterfactual, not your LLM invoice.
How does the savings share (Gainshare) work?
You pay a platform fee first. We only bill a variable share after your free savings allowance, using the same $2.50/M gateway benchmark. Search-class estimated savings count at 30% toward share; fetch, compress, dedup, pipe, and budget-block savings count at 100%. Team memory writes and Playground-only estimates do not enter the share tax base.
What is excluded from the share tax base?
Memory storage/retrieval and playground counterfactual runs are for your workflow — they are not folded into Gainshare. Share math uses audited tool savings on the six gateway modules, not your vendor LLM invoice.
Pro first month?
$5 platform fee on your first monthly Pro subscription — applied automatically at checkout.
What counts toward my token cap?
Tool token_used recorded via budget_guard and audit — including compress, fetch, search, and other MCP modules.
What is L2 hard cap (β)?
Optional hard stop when monthly cap is reached on Pro+. Alerts alone don't stop agent loops — enforcement does.
All MCP tools on Free?
Yes — full MCP tool access with a strict fair-use rate limit (30 MCP req/min per key, team ceiling 30), Playground low RPM + daily cap, and the monthly token pool. Playground is for individual evaluation, not multi-user load; bursts soft-fail with HTTP 429 (limit_scope + Retry-After). Upgrade for higher limits.
What's the difference between MCP RPM and Playground RPM?
MCP RPM is per API key for Cursor/agent integrations (Free: 30/min per key, team ceiling 30). Playground RPM is a separate team limit for Dashboard trials (Free: 10/min + daily cap). Exceeding either returns HTTP 429 — soft-fail; reduce request rate or upgrade for higher limits.
What happens when I hit a rate limit?
SmartGate returns HTTP 429 with a limit_scope (mcp_per_key, mcp_team_ceiling, or rest_team) and Retry-After when available. These are intentional fair-use entitlements; reduce request rate or upgrade for higher limits.
Is Free / Playground suitable for multi-user load?
Free / Playground is for individual evaluation with strict fair-use limits. Multi-user load may receive HTTP 429; upgrade for higher limits and shared-workspace use.
Enterprise pricing?
Contract-based. Baseline 200M+ token pool — contact sales.