Future AGI pricing

The Price List

Future AGI has two plans and three add-ons. The plans set the usage allowance and the rates above it; the add-ons buy retention, compliance, SSO and support on top of whichever plan you are on.

Both plans include 30-day retention, unlimited team members and projects, full API and SDK access, custom dashboards, heuristic evals, experiments, prompt management and the built-in guardrails. The Free plan comes with community support and needs no card; Pay-as-you-go adds email support, volume discounts, billing alerts and spending caps.

What the Meters Mean in Practice

The gateway meter only moves if you route model traffic through the Agent Command Center; tracing and evaluation alone never touch it. Storage grows with the traces and datasets you keep, so retention is a cost lever as well as a compliance setting. AI credits pay for work Future AGI's own models do. Before you estimate, ask Future AGI how many credits your evaluations will consume - that conversion is the number the rest of the budget depends on.

One line on the pricing page changes the judge bill: LLM-as-judge with your own model key (BYOK) carries a $0 platform cost. You still pay your model provider for the tokens, but the judging does not also draw down Future AGI credits.

Retention follows the add-on, not the plan: 30 days on Free and Pay-as-you-go, 90 days with Boost, a year with Scale, custom with Enterprise. A team that needs a quarter of history for incident reviews is, in practice, buying Boost.

A Worked Monthly Bill

Assume a team that, in one month, sends 2 million requests through the gateway, keeps 120 GB of data, uses 10K AI credits, runs 5 million tokens of text simulation and 500 minutes of voice simulation. Assume the list rates apply to every unit above the free allowance, with no volume discount, and that cache hits stay inside their free 100K.

Storage is the largest usage charge in this example, so filtering traces and choosing retention deliberately can materially change the bill. The add-on can also cost more than usage. A team that needs SAML SSO or a HIPAA BAA pays $750 for Scale before adding metered consumption.

Where Future AGI's Own Pages Differ

The pricing page puts SAML SSO, SCIM and a HIPAA BAA in the $750 Scale add-on with a 99.9% SLA, and lists Enterprise at $2,000 a month. The separate enterprise page presents SAML SSO, SCIM, HIPAA BAA, audit logs, ABAC and private or on-prem deployment as Enterprise features on custom pricing, with a 99.99% uptime SLA.

A standard add-on and a negotiated Enterprise contract may coexist, but the distinction affects any budget that includes SSO, a BAA or an uptime commitment. Put the applicable tier, SLA and deployment terms in the order form.

What to Ask Before You Commit

• How many AI credits does one of your evaluations consume, and does it differ between Turing Large, Turing Small and Turing Flash?
• Which SSO do you need - OAuth at Boost, or SAML with SCIM at Scale - and does your contract follow the pricing page or the enterprise page?
• Which SLA applies to you - 99.5%, 99.9% or 99.99% - and what does the financial SLA on the Enterprise add-on pay out?
• At what volume do the Pay-as-you-go discounts start?
• If you self-host, does any feature you need sit under the ee/ license?

Most of these are one email. Get the answers in writing before you size the budget, because they move the total more than the usage rates do.

What Self-Hosting Costs

The Apache-2.0 parts have no software licence fee, but self-hosting moves the cost to infrastructure and engineering. The docs size a production install at 16 or more CPU cores, 32 GB or more of RAM and 200 GB or more of SSD storage, with PostgreSQL, ClickHouse, Redis, Temporal and other services. Code in the repository's ee/ directories uses a separate licence and needs a paid subscription in production. See Future AGI self-hosting requirements for the full stack.

How It Compares with Other Published Prices

Future AGI prices by resource; most of the category prices by tier. Failproof AI, for comparison, is Free, Team at $99, Scale at $599 and Enterprise on contract, with runs and evals counted per tier and SSO/SAML arriving at Scale - see /pricing. Langfuse, Latitude, Raindrop and Galileo also publish prices; Judgment Labs does not. The units differ enough that the only honest comparison is your own volume run through each page, which is what agent observability pricing compared does.

Where Future AGI is plainly generous is seats: unlimited team members on every plan, including Free. Failproof AI's free tier is a single user.

When You Pay Nothing

A team that stays below the free storage, AI-credit, simulation and gateway allowances pays no platform usage fee. If usage grows, Pay-as-you-go adds spending caps and billing alerts. Self-hosting the Apache-2.0 parts avoids a software licence fee, but infrastructure, maintenance and any enterprise-licensed features still cost money.