API Reference
AkiliNode speaks the OpenAI API format. Any OpenAI SDK or compatible tool works out of the box.
Base URL & authentication
| Item | Value |
|---|---|
| Base URL | https://api.akilinode.com/v1 |
| Header | Authorization: Bearer sk-your-akilinode-token |
| Content type | application/json |
Create and manage tokens in the console under Tokens. A token can have its own spending cap, so one app can never drain your whole balance.
Models
| Model | Input / 1M | Output / 1M | Best for |
|---|---|---|---|
deepseek-flash | USD 0.33 | USD 1.32 | Everyday chat, code, apps — fast and cheap |
deepseek-v4-pro | USD 1.45 | USD 4.36 | Hard reasoning, analysis, complex code |
glm-5.3 | USD 1.40 | USD 4.40 | GLM flagship — reasoning, coding, agents |
qwen3.7-max | USD 2.30 | USD 6.90 | Qwen flagship for hard tasks |
glm-4.7 | USD 0.65 | USD 3.05 | Coding agents and tool use |
qwen3.7-plus | USD 0.44 | USD 1.75 | Balanced everyday model, great French |
glm-5.3-flash | USD 0.16 | USD 0.55 | Fast and cheap for apps |
qwen3.7-flash | USD 0.022 | USD 0.088 | Ultra-low-cost, high volume |
glm-4.7-flash | FREE | FREE | Free tier for testing and learning |
Cached inputs are billed at up to 98% less. List the live models any time with GET /v1/models.
Chat completions
POST /v1/chat/completions
| Parameter | Description |
|---|---|
model | Required. Any model ID from the Models table above. |
messages | Required. Array of {role, content} — roles: system, user, assistant. |
stream | Optional, default false. Set true to receive the answer token by token (server-sent events). |
temperature | Optional, 0–2. Lower = more focused, higher = more creative. |
max_tokens | Optional. Cap on the length of the reply. |
Example request
curl https://api.akilinode.com/v1/chat/completions \
-H "Authorization: Bearer sk-your-akilinode-token" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-flash",
"messages": [
{"role": "system", "content": "You are a helpful assistant."},
{"role": "user", "content": "Write a one-line slogan for a bakery in Nairobi."}
]
}'
Example response
{
"id": "chatcmpl-...",
"object": "chat.completion",
"model": "deepseek-flash",
"choices": [
{"index": 0,
"message": {"role": "assistant", "content": "Fresh from the oven, straight to your heart — Nairobi's home of warm bread."},
"finish_reason": "stop"}
],
"usage": {"prompt_tokens": 28, "completion_tokens": 19, "total_tokens": 47}
}
Billing is per token. You pay for
prompt_tokens at the input price and completion_tokens at the output price. Every call and its exact cost appear in your console under Logs.Streaming
Set "stream": true and the API returns server-sent events, one chunk at a time, ending with data: [DONE]. All OpenAI SDKs handle this with their standard streaming mode — no AkiliNode-specific code needed.
Errors
Errors come back in the OpenAI format: an error object with a message. The ones you will actually meet:
| Status | Meaning & fix |
|---|---|
| 401 | Token missing or invalid. Check the Authorization header and that the token is enabled. |
| 403 | Balance or token cap exhausted. Top up, or raise the token's limit in the console. |
| 404 | Model name not recognised. Use an exact model ID from the Models table above. |
| 429 | Too many requests in a short window. Slow down and retry with a short backoff. |
| 500 | Upstream or temporary error. Retry after a few seconds; if it persists, check Help & Support. |
© 2026 AkiliNode · Frontier AI APIs, priced for Africa