Cedar AI API

Zero-friction multilingual AI for Ghana. Translation, voice, and chat in Twi, Ewe, Ga, Dagbani, Fante, and English.

✓ No API key ✓ Streaming ✓ Voice ASR/TTS ✓ Translation

Endpoints

POST https://ai.cedar-pro.services/v1/chat/completions

POST https://ai.cedar-pro.services/api/asr

POST https://ai.cedar-pro.services/api/translate

POST https://ai.cedar-pro.services/api/tts

Languages

API Endpoints

Use Cases

Auth Required

Models

Model	Description
`cedar-ai`	Free general-purpose multilingual model. Smart, fast, supports all Ghanaian languages.
`cedar-ai-pro`	Premium model with higher accuracy. Best for translation and complex queries.
`cedar-ai-multilingual`	Optimized for multilingual conversations. Handles code-switching (mixing languages).
`cedar-ai-translate`	Dedicated translation-optimized model. Best for English to Ghanaian language translation.

Supported Languages

Code	Language	Region	Speakers
`tw`	Twi (Akan)	Ashanti, Eastern, Central	≈ 9M
`fat`	Fante	Central, Western	≈ 3M
`ee`	Ewe (Eʋegbe)	Volta	≈ 4M
`gaa`	Ga (Ģ)	Greater Accra	≈ 1M
`dag`	Dagbani (Dagbanli)	Northern	≈ 3M
`en`	English	National	All

Limits

Limit	Value
Max tokens per response	4096 (set `max_tokens` to control)
Rate limit	60 requests / minute per IP (HTTP 429 if exceeded)
Max request size	16K chars per message · 400K chars total
Authentication	None required
Max audio size (ASR)	50 MB
Max text length (Translation)	~400 tokens per request (longer text is truncated)

Request Parameters

Param	Type	Default	Required	Description
`model`	string	—	Yes	One of: `cedar-ai`, `cedar-ai-pro`, `cedar-ai-multilingual`, `cedar-ai-translate`
`messages`	array	—	Yes	Array of `{role, content}` objects
`temperature`	number	1.0	No	0–2. Lower = more precise
`max_tokens`	number	4096	No	Max response length
`stream`	bool	false	No	SSE streaming
`lang`	string	en	No	Response language: `en`, `tw`, `fat`, `ee`, `gaa`, `dag`

Language Parameter

Set lang in the request body to make Cedar AI respond in a specific Ghanaian language. The AI will reply in that language naturally.

Lang Code	Behavior
`en`	Always reply in English
`tw`	Reply in Twi (Akan)
`fat`	Reply in Fante
`ee`	Reply in Ewe (Eʋegbe)
`gaa`	Reply in Ga (Ģ)
`dag`	Reply in Dagbani (Dagbanli)

Message Roles

Role	Purpose
`system`	Sets assistant behavior, constraints, and language preference
`user`	Your questions or instructions
`assistant`	Previous AI replies for multi-turn context

Response Format

{
  // standard chat completion response
  "model": "cedar-ai",
  "choices": [{
    "message": { "content": "Akwaaba! Me din ne Cedar AI." },
    "finish_reason": "stop"
  }],
  "usage": { "total_tokens": 52 }
}

Error Codes

Code	Cause
`400`	Missing or unknown model
`413`	Message too long (max 16K chars per message, 400K total)
`429`	Rate limited — please slow down
`500`	Server error — try again
`502`	Upstream unavailable — try again shortly

Use Cases

Every example below was tested live against the API. Pick a category, copy the code, and run it.

Translation

Voice & Speech

Multilingual Chat

Agriculture

Healthcare

Education

Finance

Tourism

English → Twi Translation

curl

Tested

Translate English to Twi (Akan) using the Cedar AI LLM. Set lang to tw and ask for translation.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai-translate",
    "lang": "tw",
    "messages": [
      {"role": "system", "content": "Translate the following English text to Twi (Akan). Only respond with the translation."},
      {"role": "user", "content": "Good morning, how are you today?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai-translate",
        "lang": "tw",
        "messages": [
            {"role": "system", "content": "Translate the following English text to Twi (Akan). Only respond with the translation."},
            {"role": "user", "content": "Good morning, how are you today?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

const resp = await fetch("https://ai.cedar-pro.services/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "cedar-ai-translate",
    lang: "tw",
    messages: [
      { role: "system", content: "Translate the following English text to Twi (Akan). Only respond with the translation." },
      { role: "user", content: "Good morning, how are you today?" }
    ],
  }),
});
const data = await resp.json();
console.log(data.choices[0].message.content);

from openai import OpenAI

client = OpenAI(
    base_url="https://ai.cedar-pro.services/v1",
    api_key="x",
)

resp = client.chat.completions.create(
    model="cedar-ai-translate",
    extra_body={"lang": "tw"},
    messages=[
        {"role": "system", "content": "Translate the following English text to Twi (Akan). Only respond with the translation."},
        {"role": "user", "content": "Good morning, how are you today?"}
    ],
)
print(resp.choices[0].message.content)

Twi → English Translation

curl

Tested

Translate Twi to English. Useful when someone speaks to you in Twi and you need the English meaning.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "messages": [
      {"role": "system", "content": "Translate the following Twi to English. Only respond with the translation."},
      {"role": "user", "content": "Me pɛ sɛ me kɔ sukuu no mu."}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "messages": [
            {"role": "system", "content": "Translate the following Twi to English. Only respond with the translation."},
            {"role": "user", "content": "Me pɛ sɛ me kɔ sukuu no mu."}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

const resp = await fetch("https://ai.cedar-pro.services/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "cedar-ai",
    messages: [
      { role: "system", content: "Translate the following Twi to English. Only respond with the translation." },
      { role: "user", content: "Me pɛ sɛ me kɔ sukuu no mu." }
    ],
  }),
});
const data = await resp.json();
console.log(data.choices[0].message.content);

English → Ewe Translation

curl

Tested

Translate English to Ewe (Eʋegbe), the language of the Volta Region. Use lang: ee.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "ee",
    "messages": [
      {"role": "system", "content": "Translate the following English text to Ewe (Eʋegbe). Only respond with the translation."},
      {"role": "user", "content": "Welcome to our village celebration!"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "lang": "ee",
        "messages": [
            {"role": "system", "content": "Translate the following English text to Ewe (Eʋegbe). Only respond with the translation."},
            {"role": "user", "content": "Welcome to our village celebration!"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

const resp = await fetch("https://ai.cedar-pro.services/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "cedar-ai",
    lang: "ee",
    messages: [
      { role: "system", content: "Translate the following English text to Ewe (Eʋegbe). Only respond with the translation." },
      { role: "user", content: "Welcome to our village celebration!" }
    ],
  }),
});
const data = await resp.json();
console.log(data.choices[0].message.content);

Speech-to-Text (ASR) — Record Voice

Python

Tested

Upload an audio file (speech in Twi, Ewe, Ga, or English) and get back transcribed text. Uses Whisper ASR.

curl https://ai.cedar-pro.services/api/asr \
  -F "audio=@recording.webm" \
  -F "language=tw"

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/api/asr",
    files={"audio": open("recording.webm", "rb")},
    data={"language": "tw"},
)
print(resp.json()["text"])

const form = new FormData();
form.append("audio", audioBlob, "recording.webm");
form.append("language", "tw");

const resp = await fetch("https://ai.cedar-pro.services/api/asr", {
  method: "POST",
  body: form,
});
const data = await resp.json();
console.log(data.text);

Text-to-Speech (TTS) — Hear Responses

curl

Tested

Convert any text into spoken audio. Cedar AI will read the text aloud in a natural voice. Saves an MP3 file.

curl https://ai.cedar-pro.services/api/tts \
  -H "Content-Type: application/json" \
  -d '{"text": "Akwaaba! Mema wo akwaaba.", "language": "tw"}' \
  --output speech.mp3

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/api/tts",
    json={"text": "Akwaaba! Mema wo akwaaba.", "language": "tw"},
)
with open("speech.mp3", "wb") as f:
    f.write(resp.content)

const resp = await fetch("https://ai.cedar-pro.services/api/tts", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    text: "Akwaaba! Mema wo akwaaba.",
    language: "tw",
  }),
});
const blob = await resp.blob();
const url = URL.createObjectURL(blob);
const a = document.createElement("a");
a.href = url; a.download = "speech.mp3"; a.click();

Chat in Twi — Full Conversation

curl

Tested

Have a full conversation in Twi. Set lang: tw and Cedar AI will respond in Twi naturally. Great for building Twi-language chatbots.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai-multilingual",
    "lang": "tw",
    "messages": [
      {"role": "system", "content": "Respond in Twi (Akan) only."},
      {"role": "user", "content": "Hello! What is your name?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai-multilingual",
        "lang": "tw",
        "messages": [
            {"role": "system", "content": "Respond in Twi (Akan) only."},
            {"role": "user", "content": "Hello! What is your name?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

const resp = await fetch("https://ai.cedar-pro.services/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "cedar-ai-multilingual",
    lang: "tw",
    messages: [
      { role: "system", content: "Respond in Twi (Akan) only." },
      { role: "user", content: "Hello! What is your name?" }
    ],
  }),
});
const data = await resp.json();
console.log(data.choices[0].message.content);

from openai import OpenAI

client = OpenAI(
    base_url="https://ai.cedar-pro.services/v1",
    api_key="x",
)

resp = client.chat.completions.create(
    model="cedar-ai-multilingual",
    extra_body={"lang": "tw"},
    messages=[
        {"role": "system", "content": "Respond in Twi (Akan) only."},
        {"role": "user", "content": "Hello! What is your name?"}
    ],
)
print(resp.choices[0].message.content)

Farming Advice — Crop Disease in Twi

curl

Tested

A farmer in Ashanti can ask about crop diseases in Twi and get advice. Cedar AI understands the local context.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "tw",
    "messages": [
      {"role": "user", "content": "Me kokoo dua no ahaban redwo. Dʍn na meyʍ?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "lang": "tw",
        "messages": [
            {"role": "user", "content": "Me kokoo dua no ahaban redwo. Dʍn na meyʍ?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

Health Information — Malaria Symptoms in Ewe

curl

Tested

A patient in Volta Region can ask about malaria symptoms in Ewe. Cedar AI provides health information in their mother tongue.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "ee",
    "messages": [
      {"role": "user", "content": "Nye dɔléle dɔ asi le yem. Azɔ?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "lang": "ee",
        "messages": [
            {"role": "user", "content": "Nye dɔléle dɔ asi le yem. Azɔ?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

Math Tutoring in Ga

curl

Tested

Students in Greater Accra can get math tutoring in Ga. Cedar AI explains concepts in their first language.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "gaa",
    "messages": [
      {"role": "user", "content": "Minawo lɛ nɔ, 2+2 ya?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "lang": "gaa",
        "messages": [
            {"role": "user", "content": "Minawo lɛ nɔ, 2+2 ya?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

Mobile Money Help in Dagbani

curl

Tested

Users in Northern Region can ask about mobile money (MoMo) in Dagbani. Cedar AI helps with financial literacy.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "dag",
    "messages": [
      {"role": "user", "content": "N yi bɔri ni n tum liɣiri mobile money zuɣu, bo n-nyɛ dini?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "lang": "dag",
        "messages": [
            {"role": "user", "content": "N yi bɔri ni n tum liɣiri mobile money zuɣu, bo n-nyɛ dini?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

Tourist Guide — Kakum Park in Fante

curl

Tested

A visitor to Central Region can ask about Kakum National Park in Fante. Cedar AI provides tourism info.

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "fat",
    "messages": [
      {"role": "user", "content": "Kakum National Park no ho nsʍm ka kyerew me?"}
    ]
  }'

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={
        "model": "cedar-ai",
        "lang": "fat",
        "messages": [
            {"role": "user", "content": "Kakum National Park no ho nsʍm ka kyerew me?"}
        ]
    },
)
print(resp.json()["choices"][0]["message"]["content"])

LLM Chat API

POSThttps://ai.cedar-pro.services/v1/chat/completions

OpenAI-compatible chat completions. No API key required. Supports streaming, multilingual responses, and all standard parameters.

Speech-to-Text API

POSThttps://ai.cedar-pro.services/api/asr

Upload audio files (webm, wav, mp3, m4a) and get transcribed text. Supports Twi, Ewe, Ga, Dagbani, and English.

Parameter	Type	Required	Description
`audio`	file	Yes	Audio file (webm, wav, mp3, m4a). Max 50MB.
`language`	string	No	Hint: `en`, `tw`, `fat`, `ee`, `gaa`, `dag`

Translation API

POSThttps://ai.cedar-pro.services/api/translate

Translate text between English and any Ghanaian language using NLLB-200.

Parameter	Type	Required	Description
`text`	string	Yes	Text to translate
`source_lang`	string	No	Source language code (default: `en`)
`target_lang`	string	No	Target language code (default: `tw`)

Text-to-Speech API

POSThttps://ai.cedar-pro.services/api/tts

Convert text to natural-sounding speech audio (MP3 format).

Parameter	Type	Required	Description
`text`	string	Yes	Text to convert to speech
`language`	string	No	Language code (default: `en`)

Note All languages are currently voiced with a clear multilingual speech engine that reads Ghanaian-language text phonetically. Native-accent voices for Twi, Ewe, Ga and Dagbani are on the roadmap.

cURL

curl https://ai.cedar-pro.services/v1/chat/completions \
  -H "Content-Type: application/json" \
  -d '{
    "model": "cedar-ai",
    "lang": "tw",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'

Python

import requests

resp = requests.post(
    "https://ai.cedar-pro.services/v1/chat/completions",
    json={"model": "cedar-ai", "lang": "tw", "messages": [{"role": "user", "content": "Hello!"}]},
)
print(resp.json()["choices"][0]["message"]["content"])

JavaScript (Fetch)

const resp = await fetch("https://ai.cedar-pro.services/v1/chat/completions", {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "cedar-ai",
    lang: "tw",
    messages: [{ role: "user", content: "Hello!" }],
  }),
});
const data = await resp.json();
console.log(data.choices[0].message.content);

OpenAI Python SDK

from openai import OpenAI

client = OpenAI(
    base_url="https://ai.cedar-pro.services/v1",
    api_key="x",
)

resp = client.chat.completions.create(
    model="cedar-ai",
    extra_body={"lang": "tw"},
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)

Tip Use the OpenAI Python SDK — just set base_url to our endpoint and anything as api_key. Add extra_body={"lang": "tw"} to get responses in Ghanaian languages.