Pricing

Predictable licensing.

Flat per-seat pricing. Volume discounts at scale. We don't mark up compute, and we don't arbitrage LLM API calls. Deploy our free, two-week pilot to prove the value before committing.

Three tiers. What's in each.

Connect
Start here
$18 / seat / mo
Billed monthly or annually
  • Connect proxy + workspace daemon per developer machine
  • VS Code extension
  • Single-tenant Overwatch in your cloud account
  • Deterministic regex floor + operator exceptions
  • CONNECT-tier marketplace blocking
  • Configured-upstream content inspection (Tier B)
  • Optional LLM judge — customer-supplied (your cloud model or a local one)
  • JIT approval queue + append-only audit
  • MCP tether_may_i_* voluntary surface
Overwatch Enterprise
Org-wide add-on
$8 / seat / mo + $2,500 / org / mo base
Required for orgs ≥ 500 seats
  • SSO via SAML 2.0 / OIDC (Okta, Entra ID, Auth0, Google Workspace)
  • SCIM directory sync via Jackson
  • Long-term audit retention (90d Connect / 1y Workspace / 7y Enterprise)
  • Evidence your auditor maps to your SOC 2 / FedRAMP program
  • Custom branding bundle distribution

The discounts — prepay, lighthouse, early-stage — are all listed once in the Discounts at a glance block below. Prices reflect the current pricing spec.

The bigger the team,
the lower the seat price.

Volume pricing tiers for Connect and Overwatch Enterprise base. Workspace volume tiers withheld during preview.
SeatsConnectOverwatch base
1 – 49$18 / seat / mo
50 – 249$15 / seat / mo$1,500 / mo
250 – 999$12 / seat / mo$2,500 / mo
1000+$10 / seat / mo$4,000 / mo (custom features available)

Workspace volume tiers are withheld until GA — pricing on request during preview. Education and qualified non-profits: 50% off list (see Discounts at a glance).

Worth it? One customer record or cloud key that leaks through an AI tool can cost far more than a year of Tether. And the judge/inference runs on a model you bring — so there's no extra per-request AI cost from us.

One source of truth.
Listed once.

Discounts are not stackable unless we say so in writing. The FAQ and SKU footnotes point here.

  • Annual prepay15% off list.
  • 3-year prepay25% off list.
  • 5-year prepay35% off list. Rare.
  • LighthouseFirst 5 paid customers in any vertical: 50% off for 12 months in exchange for a public case study and reference calls.
  • Early-stage startups<$5M ARR and <25 seats: Connect free for 12 months.
  • Education / non-profit50% off list for qualified accounts.

Two-week pilot.
Free up to 25 seats.

The Connect tier, in your own cloud, for two weeks at no cost. On day 14 you decide: keep going at list price, or run terraform destroy — there's nothing on our side to delete. Prepay and lighthouse terms are in the Discounts at a glance block.

The Workspace tier in pilot is for design partners only, on per-deal terms. The full scope — what's included, the day-by-day plan, success criteria, SLA, and the limits up front — lives on the Pilot page instead of being repeated here.

Best-effort SLA during pilot. Production SLA post-GA.

Everything you can verify in the two weeks — and the limits we name up front — is laid out on the Pilot page.

What the price does
and does not include.

Is judge inference included?

You bring your own model for the judge. A cloud judge bills from your own model vendor — Tether charges nothing for judge calls. A local judge runs on the developer's machine or a server you host, at zero cost per call. You set how often the judge samples, so the cost stays predictable.

Why is the judge cost not on the bill?

Model prices vary by 100x across vendors. Rolling them into one per-seat number would throw off the math for half our customers. At the default sample rate, judge calls are tiny next to the model spend your developers already run up — and $0 on a local judge. Ask, and we'll give you a concrete range for your seat count and judge model.

Are custom integrations included?

Your first three SAML or SCIM connector setups are included. After that, each custom integration is a flat $2,500 per connector. Special data-residency needs — EU-only Overwatch, on-prem only — add 30% to base list.

What's the cloud infrastructure cost?

You run one small per-tenant service — Cloud Run on GCP or ECS Fargate on AWS — and pay your cloud vendor directly. Plan for roughly a few tens of dollars a month, depending on cloud and traffic. It stays predictable because the service bills by instance-hour, not per request, so the bill is flat per tenant. We'll size it for your environment on request.

What is the SLA?

Best-effort SLA during pilot. Production SLA post-GA.

What discount terms exist?

One source of truth: the Discounts at a glance block above. Annual / 3-year / 5-year prepay, lighthouse, early-stage, education — all listed once.

Bring your seat count.
We'll bring the math.

Discounts start at 50 seats. Lighthouse terms are on the table. The floor is $10 a seat at 1000 and up.

Book a 30-min walkthrough ↗