Skip to content

Details

Before you start

Make sure your account is ready for inference:

  • Account has balance.
  • Public key is published on-chain (inferenced publish-pubkey for CLI-created accounts, or any on-chain transaction for external wallets).
  • GET /v2/accounts/:address returns your account.
  • You do not need Participant registration for developer inference.

For setup steps, see Developer Quickstart.

Inference Request with inferenced

Create a Directory for Request Payloads 

mkdir inference-requests

Save the payload for an OpenAI-compatible /chat/completions request in a file inside the inference-requests directory. For example, create a file named inference-requests/request_payload.json with the following content:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
{
  "model": "Qwen/Qwen3-235B-A22B-Instruct-2507-FP8",
  "messages": [
    {
      "role": "user",
      "content": "What is the capital city of France?"
    }
  ],
  "stream": false
}

Run the following command to submit your inference request:

./inferenced signature send-request \
  --account-address $GONKA_ADDRESS \
  --node-address $NODE_URL \
  --file ./inference-requests/request_payload.json

Using the Gonka API directly

To access Gonka API directly you have to: 1. Use of the randomly selected endpoint 2. Signs the request body with your private key using ECDSA 3. Adds the signature to the Authorization header 4. Add your account address to the X-Requester-Address header

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
import os
import json
import time
import requests
import hashlib
import base64
from ecdsa import SigningKey, SECP256k1

# Required environment variables
# GONKA_PRIVATE_KEY: your hex private key (with or without 0x)
# GONKA_ADDRESS: your gonka address (bech32) used in X-Requester-Address
# GONKA_ENDPOINT: the endpoint base URL INCLUDING /v1 (e.g. https://host:port/v1)
# GONKA_PROVIDER_ADDRESS: provider's gonka address for this endpoint (used in signature)

GONKA_PRIVATE_KEY = os.environ['GONKA_PRIVATE_KEY']
GONKA_ADDRESS = os.environ['GONKA_ADDRESS']
GONKA_ENDPOINT = os.environ['GONKA_ENDPOINT']
GONKA_PROVIDER_ADDRESS = os.environ['GONKA_PROVIDER_ADDRESS']

def sign_payload(payload_bytes: bytes, private_key_hex: str, timestamp_ns: int, provider_address: str) -> str:
    # Remove 0x prefix if present
    pk = private_key_hex[2:] if private_key_hex.startswith('0x') else private_key_hex
    sk = SigningKey.from_string(bytes.fromhex(pk), curve=SECP256k1)

    # Message bytes: payload || timestamp || provider_address
    msg = payload_bytes + str(timestamp_ns).encode('utf-8') + provider_address.encode('utf-8')

    # Deterministic ECDSA over SHA-256 with low-S normalization
    sig = sk.sign_deterministic(msg, hashfunc=hashlib.sha256)
    r, s = sig[:32], sig[32:]
    order = SECP256k1.order
    s_int = int.from_bytes(s, 'big')
    if s_int > order // 2:
        s_int = order - s_int
        s = s_int.to_bytes(32, 'big')
    return base64.b64encode(r + s).decode('utf-8')

# Hybrid timestamp (monotonic + aligned to wall clock) like the library
_WALL_BASE = time.time_ns()
_PERF_BASE = time.perf_counter_ns()
def hybrid_timestamp_ns() -> int:
    return _WALL_BASE + (time.perf_counter_ns() - _PERF_BASE)

payload = {
  "model": "Qwen/Qwen3-235B-A22B-Instruct-2507-FP8",
  "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, how are you?"}
  ],
  "stream": True,
}

payload_bytes = json.dumps(payload).encode('utf-8')
timestamp_ns = hybrid_timestamp_ns()
signature = sign_payload(payload_bytes, GONKA_PRIVATE_KEY, timestamp_ns, GONKA_PROVIDER_ADDRESS)

response = requests.post(
  url=f"{GONKA_ENDPOINT}/chat/completions",
  headers={
    "Content-Type": "application/json",
    "Authorization": signature,
    "X-Requester-Address": GONKA_ADDRESS,
    "X-Timestamp": str(timestamp_ns),
  },
  data=payload_bytes,
)

response.raise_for_status()
response_json = response.json()
assistant_message = response_json['choices'][0]['message']['content']

print("\nAssistant says:", assistant_message)