APIOpen console

Small-model utility APIs.

Speech, vectors, voice, and OCR through one OpenAI-compatible gateway with clear dollar pricing.

Open consoleView pricing
Abstract model routing infrastructure for speech, embeddings, text to speech, and OCR
Gateway route
API keyNew APIWorkers
asr-smallembed-smalltts-smallocr-small

One interface for auxiliary models.

Use one API account for ASR, embeddings, TTS, and unlimited OCR. Each model keeps a plain endpoint and a small public name.

ASR

Speech-to-text for clips, support calls, meeting snippets, and voice notes.

/v1/audio/transcriptions

Embeddings

Vector generation for search, recommendations, RAG memory, clustering, and routing.

/v1/embeddings

TTS

Text-to-speech for product narration, agents, voice replies, and audio jobs.

/v1/audio/speech

Unlimited OCR

Document and screenshot extraction for high-volume visual parsing.

/v1/chat/completions

Dollar pricing that can be audited.

Built for US and European buyers. Input, output, and cached input are tracked as separate billing dimensions.

Input$0.10

per 1M billable input tokens

Output$0.50

per 1M generated output tokens

Cache hits$0.00

prompt-cache hits are free

Local runtime, public API gateway.

happytoken.app serves the public site. api.happytoken.app routes to New API, then to local worker channels.

One OpenAI-compatible surface

User keys, channels, quotas, logs, model names, and billing stay in one New API control plane.

Local llama.cpp workers

Small utility models run on local worker endpoints behind the same gateway and domain plan.

Dollar metering

Input, output, and cache usage are metered separately so each request can be audited.

OpenAI-compatible request
curl https://api.happytoken.app/v1/embeddings \
  -H "Authorization: Bearer $HAPPYTOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "embed-small",
    "input": "Index this support ticket for semantic search."
  }'

Ready for repeatable model work.

Sign in, add credit, create an API key, and call the same gateway for small-model workloads without unrelated product flows.

Open console