Simple, transparent pricing.

Start free. Upgrade when you're ready. No surprises.

Free

For solo devs and first experiments.

$0forever
How it's priced
Base$0 / mo
SeatsUnlimited seats, $0
Runs5M incl, hard cap (no overage)
  • 10 active experiments
  • Unlimited feature flags
  • mSPRT & T-Test
  • 1 workspace
  • 1-year data retention
  • 7-day audit log
  • 200 prompt merges (Construct) / month
  • Sidekick AI assistant: 14-day trial, then 50 msgs/mo
Start Free →

Work email + card on file required at signup.

Most Popular

Pro

For teams running experiments in production.

$99/ seat / mo
How it's priced
Base$0 / mo
Seats$99 / seat
Runs25M incl, $0.03 / 1K over
  • AI optimization agent + Sidekick (1,500 msgs/mo)
  • Unlimited experiments
  • Runtime guardrails & kill switches
  • Warehouse-grade analytics (100M rows)
  • All statistical methods
  • Integrations (Slack, Datadog, webhooks)
  • 60-day audit log
  • 5 workspaces, 1-year retention
  • Email support (48hr SLA)
Start Free →

Team

For growing eng teams that want flat-rate billing.

$349/ mo
How it's priced
Base$349 / mo
Seats10 incl, +$30 / additional
Runs100M incl, $0.02 / 1K over
  • Everything in Pro
  • Sidekick: 5,000 msgs/mo pooled
  • Multi-agent tracing
  • Advanced runtime guardrails
  • SSO / SAML + RBAC
  • 90-day audit log
  • Warehouse-grade analytics (1B rows)
Start Free →

Included on every plan

Feature flags (unlimited on all plans)
Dynamic configs
User segments
Prompt versioning
Full SDK access (Python, Node, Go)
REST API access

Need more?

Scale adds volume and priority support; Enterprise adds self-hosting and a reliability SLA.

Scale

For higher volume and priority support.

$1,499/ mo
How it's priced
Base$1,499 / mo
Seats20 incl, +$75 / additional
Runs250M incl, $0.01 / 1K over
  • Everything in Team
  • Sidekick: 20,000 msgs/mo pooled
  • 2-year data retention
  • 180-day audit log
  • Priority support (2hr SLA)
Start Free →

Enterprise

For regulated or self-hosted deployments.

Custom
  • Self-hosted / VPC deployment
  • 99.9% uptime SLA
  • Dedicated CSM
  • Negotiated event volume
  • Sidekick: bring your own Anthropic key + custom personalities
Contact Sales

Estimate your cost

Most teams start on Free and upgrade as they grow.

Solo developer

1 seat · < 5M runs/mo
Base$0
Seats · 1 × $0$0
Usage · 5M incl · hard cap$0
Free
$0 /mo

Team of 5

5 seats · 5–25M runs/mo
Base · 10 seats + 100M included$349
Seats · +0 × $30$0
Usage · 100M incl · $0.02/1K over$0
Team
$349 /mo

Team of 12

12 seats · ~80M runs/mo
Base · 10 seats + 100M included$349
Seats · +2 × $30$60
Usage · 100M incl · $0.02/1K over$0
Team
$409 /mo

Team of 25

25 seats · 100–250M runs/mo
Base · 10 seats + 100M included$349
Seats · +15 × $30$450
Usage · 100M incl · $0.02/1K over$0
Team
$799 /mo
Above ~154M runs Scale becomes cheaper.

Frequently asked questions

Do I need a credit card to start?

Yes. The Free plan requires a credit card and a work email at signup. Free-mail addresses (gmail, yahoo, outlook, etc.) are rejected. You will not be charged unless you upgrade or exceed Free-tier limits, but requiring a card up-front keeps abuse and throwaway accounts out of the product.

What counts as an agent run?

An agent run is any observation sent through the SDK: experiment exposures, metric logs, or eval results. Feature flag evaluations on Free are unlimited and do not count.

Can I switch plans anytime?

Yes. Upgrade or downgrade at any time. When upgrading, you get immediate access to new features. When downgrading, changes take effect at the next billing cycle.

What happens if I exceed my agent-run limit?

Free is hard-capped at 5M agent runs/month — once you hit the cap, ingest returns HTTP 402 until next month or until you upgrade. Paid plans are uncapped: overage bills at $0.03/1K (Pro), $0.02/1K (Team), or $0.01/1K (Scale) at the end of the billing cycle, with warnings at 80% and 100% so the bill is never a surprise.

Is there an annual discount?

Yes. Annual billing saves ~10% on Pro and Scale plans. Contact us for Enterprise annual pricing.

Can I self-host Winnow?

Yes. Self-hosted deployment is part of Enterprise. The core runs on SQLite with no required external dependencies.

When does warehouse-grade analytics start to matter?

Warehouse-grade column-store analytics is available on every plan. In practice it becomes meaningful at Pro volumes (25M+ agent runs/month) where column-store aggregates beat SQLite for dashboard queries. Row caps scale with tier: 10M on Free, 100M on Pro, 1B on Team, 5B on Scale, unlimited on Enterprise.

What is the Sidekick?

The Sidekick is an in-app AI assistant that helps you read and steer the platform: it explains scorecards, summarizes what the optimization agent is doing, runs cookbooks, and drafts experiments. It only ever proposes changes; every change to production still passes a human approval. You pick a personality once and the same assistant answers in the app and in your IDE over MCP.

What counts as a Sidekick message?

One message is one turn you send the Sidekick (a question or an instruction). Reading the answer and the follow-up tool calls it makes do not each cost a message. Free includes a 14-day full trial and then 50 messages a month; Pro includes 1,500, Team 5,000 pooled, Scale 20,000 pooled, and Enterprise brings its own Anthropic key.

What happens when I run out of Sidekick messages?

Nothing breaks. The Sidekick tells you that you are out for the month and still gives you the free read-only help (doc lookups, deep links, and quick actions that do not call the model). Upgrade for a larger monthly allowance.