§ Text API
Text API. 4 endpoints, $0.001 a call.
Two pairs of endpoints for two different jobs. Redact and detect throw information away and nothing puts it back. Tokenize and restore are the round trip you put either side of a model call, so the provider never receives the personal information and your user still gets a useful answer. Runs on go-phileas.
curl -sS --fail-with-body \
-o redacted.txt \
-H "Authorization: Bearer $OHMYWRAP_KEY" \
-F "file=@notes.txt" \
-F 'options={"detect":["email-address","ssn"],"strategy":"redact"}' \
https://api.ohmywrap.com/text/redactThe mapping never leaves the caller.
Tokenize returns the map in its response and we keep nothing, so the round trip needs no session, no job id, and no coordination between the two calls — they can land on different workers. The values in that map were already yours; handing them back retains nothing. A round trip is two calls, which is $0.002.
- Bytes per file
- 2 MB
- Bytes per request
- 4 MB
- Inline dictionary terms
- 1,000
- Bytes per term
- 256
- Distinct tokens per request
- 5,000
- Encoding
- UTF-8 only
- Job time budget
- 30s
- Price per call
- $0.001
Anything over a limit is refused immediately, and refusals are not billed.
§ The inventory
Every text endpoint.
Each is a POST with your file attached, each returns the result on the same connection, and each costs $0.001. Follow any of them for parameters, limits, and a request you can paste.
Find and remove
Seventeen detectors, allowlisted, of which thirteen are on by default — date, age, zip-code and url are available but off, because they fire constantly in ordinary prose. Callers never supply an engine policy: they supply a closed options object and the service builds the policy, because a go-phileas policy is a program and three of its features reach outside the process. Named entities — people, organisations, places — are not offered for exactly that reason: the engine finds them through an external HTTP service, which is the one thing that boundary exists to keep unreachable. Detection is regular expressions and an exact-match term list. It will miss a name in running prose. Use detect to find out whether a document needs a human, not to certify that it does not.
Your first text call takes one line of curl.
No plan to choose. No credits to model. No storage to configure.
10,000 calls for $10. No subscription, no expiry.