$0/mo
Try it on your own calls
- 50K gateway calls · 1 compile credit · 1 seat
- Local CLI and open capability JSON.
- Point a few calls at Kolm and watch it capture.
- Compile one model and see what you'd own.
PRICING · STOP RENTING
Pay to run your model, not to rent someone else's.
Once Kolm captures how you use the API, you compile that behavior into a signed model and run it on your own hardware. Every call after that is yours - not a frontier bill you pay forever to re-run work you've already done. Six plans, real limits, no surprises. Every plan, with the exact numbers, lives at GET /v1/plans.
6 plans, free to enterprise
25M Business gateway calls
200 Business compile credits
17% off when you pay yearly
- PRICE
- $49/ mo
- GATEWAY
- 500Kcalls
- COMPILE
- 50credits
- SEATS
- 1seat
PLANS · GET /V1/PLANS
Every plan, with real limits.
Pick where you are today and grow when you need to. These are the same plans served at GET /v1/plans; the numbers you see here are the numbers you get. Start a workspace and you're provisioned right away, with billing you can read in full at any time. Two meters do the work: gateway calls, the live API requests Kolm captures, and compile credits (one credit builds one model)A compile credit is one model build. Spend one credit and Kolm turns your captured traffic into a single portable .kolm file: model, recipe, evals, manifest, and signed receipt. Routing your live calls spends gateway calls; building a model from them spends a compile credit..
$29/mo
Ship your first owned model
- 500K gateway calls · 10 compile credits · 1 seat
- Capture your live traffic and preview the model first.
- Compile to a portable file and export the receipt.
- 17% off yearly · built for the one-person production app.
$49/mo
Move real workloads onto your hardware
- 500K gateway calls · 50 compile credits · 1 seat
- Route across providers, export captures, download your models.
- See which hardware fits before you deploy - with the limits shown.
- 17% off yearly · for teams turning repeat API work into models they own.
$99/mo
Build and review models together
- 5M gateway calls · 50 compile credits · 5 seats
- Five seats with shared workspaces and access controls.
- Combine specialist models and review them as a team.
- 17% off yearly · for teams shipping models side by side.
$499/mo
Run owned models across a fleet
- 25M gateway calls · 200 compile credits · 20 seats
- High call caps with role-based control over who captures what.
- Roll models out to a device fleet or let us host the serve path.
- 17% off yearly · for production programs on devices and hosted runtimes.
Custom
Own it in your own cloud
- 250M gateway calls · custom compile credits · 25 seats base
- Deploy in your private cloud, with compile credits sized to you.
- Hands-on rollout, plus the procurement packet your buyers need.
- Talk to us · for private-cloud, regulated, and high-volume programs.
Want the raw numbers? Every plan is published at /v1/plans and /v1/billing/tiers - the rates are public and there are no hidden line items.
COMPARE · QUOTAS
Compare every plan, side by side.
You pay for the work that builds something you keep - the API traffic you capture, the models you compile, your seats. No filler line items. The numbers below come straight from the plans above.
| Plan | Gateway calls / mo | Compile credits / mo | Seats | Price |
|---|---|---|---|---|
| Free | 50K | 1 | 1 | $0/mo |
| Indie | 500K | 10 | 1 | $29/mo |
| Pro | 500K | 50 | 1 | $49/mo |
| Team | 5M | 50 | 5 | $99/mo |
| Business | 25M | 200 | 20 | $499/mo |
| Enterprise | 250M | Custom | 25 base | Custom |
FIT · WHO EACH PLAN IS FOR
Which plan, and what you get in it.
Same two meters across every plan; the difference is scale and who is in the workspace. The matrix marks each capability with a word, not just a tint, so it scans without color. Then a worked example shows how the meters add up on a real month, using only the prices already on this page.
| Capability | Free | Indie | Pro | Team | Business |
|---|---|---|---|---|---|
| Capture live traffic | try | yes | yes | yes | yes |
| Compile a model you own | 1 | 10 | 50 | 50 | 200 |
| Export and download artifacts | no | yes | yes | yes | yes |
| Shared workspace and seats | 1 | 1 | 1 | 5 | 20 |
| Role-based access control | no | no | no | team | yes |
| Fleet rollout or hosted serve | no | no | self | self | yes |
TWO METERS gateway calls + compile credits same at /v1/plans
Worked example · one month on Pro
- Pro plan$49/mo
- Gateway calls used (within 500K cap)included
- Models compiled this month (within 50 credits)included
- Total this month$49/mo
Worked from the prices already listed above. On Pro, up to 500K captured calls and up to 50 model builds are inside the flat $49/mo, so a month that stays under both caps is just the plan price. Cross a cap and you step up to the next plan whose numbers fit; nothing meters past the limit silently. Move to yearly and the same plan is 17% off.
Read it in one line.
Solo and proving it on your own calls: Free. One production app, one owner: Indie at $29/mo. Real workloads you want to move onto your own hardware: Pro at $49/mo. A team building and reviewing models together: Team at $99/mo. A fleet running owned models in production: Business at $499/mo. Private cloud, regulated, or high volume: Enterprise. Two meters decide it: how many calls you capture, and how many models you compile.
BILLING · CONTRACT
See exactly what each charge covers.
No mystery line items. Every meter maps to one route in the same backend that powers your signup - so you always know what you're paying for and where it happens.
| Meter | What it covers | Route |
|---|---|---|
| Gateway | Your live OpenAI- and Anthropic-compatible calls, captured as they run - with routing, caching, and fallback so traffic keeps flowing. | /v1/route/chat/completions |
| Estimate | Tell us your call volume, compile work, seats, SSO, and where you'll deploy - we recommend the right plan without touching your secrets. | /v1/pricing/estimate |
| Compile | Turn captured behavior into a portable model - preview it, build it, run the evals, and package the manifest and signed receipt. | /v1/compile |
| Control | Set the rules in one place - sources, provider policy, credentials, retention, redaction, and exports - and see what's ready to ship. | /v1/account/api-control-center |
| Runtime | Find the smallest hardware that fits, deploy there, and track installs - or let us host the serve path. Export the model and run it anywhere. | /v1/devices |
FAQ · WHAT YOU PAY FOR
What your money buys.
One plan takes you the whole way - from your live API calls to a portable, signed model you run on your own hardware. Here's how the pieces price out.
Which plan is right for me?
Tell us your call volume, compile work, seats, SSO, and where you'll deploy, and GET /v1/pricing/estimate points you at a plan - the same plans you see here, so the estimate matches your signup.
Why are compile credits their own line?
Because building a model is the part that creates something you own - so you can see exactly what you pay for it, never buried in infrastructure. One credit builds one model: examples, recipe, evals, manifest, and signed receipt.
What comes with the bigger plans?
Seats, role-based access, shared workspaces, provider policy, and exports - real controls wired into the product, not generic add-ons stapled onto a seat count.
Can I take my work with me?
Yes - it's the whole point. Your models, manifests, receipts, reports, and procurement packets all export and work outside our UI. No lock-in. Even the plan data is open at /v1/plans and /v1/billing/tiers.
CTA · START FREE
Ready to own what you're renting?
Point a few calls at Kolm, watch it capture, compile one model, and check the receipt yourself. Prove it on one workload, then scale when you're ready - the Free plan costs nothing and needs no card.