Brioworkx AI Platform

Your API base URL

Available models

brioworkx-chat — chat / completions (Qwen3-Next-80B).
bge-m3 — text embeddings (1024-dim, multilingual).
rerank-8b — relevance reranking (Qwen3-Reranker-8B, Cohere-style).
whisper-stt — speech-to-text (faster-whisper large-v3-turbo, multilingual).
TTS — text-to-speech (OmniVoice, voice cloning, 600+ languages). Endpoint: /v1/tts, /v1/tts/languages.
OCR — text extraction (RapidOCR). Endpoint: /v1/ocr/image, /v1/ocr/file.
Guard — safety moderation (Qwen3Guard, multilingual). Endpoint: /v1/moderate.

Get a key

1. Open the Console and sign in (admin credentials).
2. Go to Virtual Keys → Create New Key, scope it to the models you need (brioworkx-chat, bge-m3, rerank-8b, whisper-stt). The same key also authorizes the OCR endpoints.
3. Copy the sk-… key — you'll see it once.

curl https://ai-platform.fincover.com/v1/chat/completions \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"brioworkx-chat","messages":[{"role":"user","content":"Hello"}]}'

from openai import OpenAI
client = OpenAI(base_url="https://ai-platform.fincover.com/v1",
                api_key="YOUR_BRIOWORKX_KEY")
r = client.chat.completions.create(
    model="brioworkx-chat",
    messages=[{"role": "user", "content": "Hello"}])
print(r.choices[0].message.content)

curl https://ai-platform.fincover.com/v1/embeddings \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"bge-m3","input":"hello world"}'

Returns a 1024-dim vector. Same shape as OpenAI /v1/embeddings — works with the OpenAI SDK via client.embeddings.create(model="bge-m3", input=…).

curl https://ai-platform.fincover.com/v1/rerank \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"rerank-8b","query":"waiting period for cataract","documents":["Cataract has a 24-month waiting period.","The free-look period is 15 days."],"top_n":2}'

Cohere-style rerank — returns results sorted by relevance_score (each with the original index).

curl https://ai-platform.fincover.com/v1/audio/transcriptions \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -F "model=whisper-stt" \
  -F "file=@audio.wav"

OpenAI Audio API shape — returns {"text": …}. Language auto-detects; pass -F "language=ta" to force one. Works with the OpenAI SDK via client.audio.transcriptions.create(...).

# file upload
curl https://ai-platform.fincover.com/v1/ocr/file \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -F "file=@page.png"

# or base64 JSON
curl https://ai-platform.fincover.com/v1/ocr/image \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"ocr","image":"<base64-or-data-url>"}'

Returns {"text", "lines":[{text, score}]}. Brioworkx extension (not part of the OpenAI spec) — same base URL & key. Include "model":"ocr" so usage shows up under ocr (not unknown) in the dashboard.

curl https://ai-platform.fincover.com/v1/tts \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input":"Hello from Brioworkx","voice":"nirva","language":"English"}' --output speech.wav

# list the 600+ supported languages
curl https://ai-platform.fincover.com/v1/tts/languages \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY"

OmniVoice zero-shot voice cloning. voice = a cloned persona; language = any of 600+ (or omit / "Auto" to auto-detect) — the same voice speaks them all. Returns a WAV.

curl https://ai-platform.fincover.com/v1/moderate \
  -H "Authorization: Bearer $BRIOWORKX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"prompt":"...user text...","response":"...optional assistant text..."}'

Qwen3Guard (multilingual). Returns {"safety":"Safe|Controversial|Unsafe","categories":[…],"refusal":…}. Callable — not auto-enforced on your chat calls.