ONE WORK EMAIL - NO SDK REWRITE
Get a key. Own the model.
Point your existing OpenAI and Anthropic calls at Kolm - no rewrite. It captures how you actually use the API, compiles it into a signed model in one portable file, and runs it on your own hardware. Stop paying frontier prices to re-run behavior you've already established.
01 · GET A KEY
Start with a key, not a sales maze.
Self-serve and fast: one email gets your key, you point one namespace (one named slice of your API traffic)A namespace is a named slice of your API traffic, for example one project, one route, or one environment. You can point just one at Kolm to try it, and leave the rest of your stack untouched. at Kolm, and you watch your captured traffic turn into a signed model in the API Control Center. Talk to sales only when you want to.
- Endpoint
- POST /v1/signup
- Returns
- api_key, tenant metadata, onboarding guidance, and billing state when relevant.
- Session
- The server sets kolm_session so your account opens without pasting the key.
Get your API key
Use a work email. You start on the Free tier, no card needed. Click once and your key appears on this page right away, shown a single time, so copy it before you leave. After that you can capture traffic, set provider rules, run evals, compile a signed model, and export it, all yours to keep.
Example: you@company.com. New here? Get a key. Returning? Email yourself a one-time sign-in link.
By creating an account you agree to the Terms and Privacy Policy. Already set up? Open the API Control Center.
You're in. Here's your key.
Workspace active on Free.
- 1Point one namespace at Kolm with your key - keep your SDK exactly as it is.
- 2Capture enough real traffic to build your eval and regression sets.
- 3Compile it into a signed .kolm model you own and run on your own hardware.
Prove it on one route before you move anything.
FIRST 10 MINUTES
You don't migrate your stack to find out if this works. Point one route at Kolm and watch it go end to end - key, capture, compile, signed model - before you commit anything else.
- ·Get a scoped key · keep it in account controls, not scattered across env files.
- ·Point one OpenAI-compatible namespace at Kolm before any broad migration.
- ·Capture the prompts, completions, tool calls, labels, and failures you actually run.
- ·Get back a signed model - manifest, eval evidence, runtime target, receipt - that's yours.
See every key, route, and signed model in one place.
ENTERPRISE CONTROLS
Your team owns the bill and the trust story, so you get the full picture - not just the model call. Every route maps to the same objects, and every change is something you can inspect or export at any time.
- Identity
- workspace, project, environment, credential, key scope, user role, account lifecycle.
- Sources
- API routes, SDKs, CLI uploads, webhooks, streams, logs, traces, browser events, warehouse imports, labels, receipts.
- Policy
- model allowlists, connector state, redaction, retention, rate limits, cost controls, risky-action approvals.
- Evidence
- receipts, manifests, eval reports, event streams, warehouse exports, webhook callbacks, runtime packages, readiness ledgers.
curl -s https://kolm.ai/v1/signup \
-H "Content-Type: application/json" \
-d '{"email":"you@company.com"}'
curl -s https://kolm.ai/v1/route/chat/completions \
-H "Authorization: Bearer $KOLM_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"namespace":"default",
"model":"openai:gpt-4.1-mini",
"messages":[{"role":"user","content":"hello"}]
}'
Ready to own what you're renting?
One email gets you a key. Point one route, capture your real traffic, and walk away with a signed model you run on your own hardware - no rewrite, no lock-in.