1. Settings (Ctrl/Cmd+,) → Models → + Add Custom Model
2. Model name: kr/claude-sonnet-4.6
3. Scroll ke OpenAI API Key → centang Override OpenAI Base URL
4. Base URL: https://api.wflab.web.id/v1
5. API Key: paste sk-xxx
6. Klik Verify → muncul checkmark
7. Tambah model lain dengan repeat step 2 (cuma ganti model name)

Edit ~/.continue/config.json:

{
  "models": [
    {
      "title": "Sonnet 4.6 (cheap)",
      "provider": "openai",
      "model": "kr/claude-sonnet-4.6",
      "apiBase": "https://api.wflab.web.id/v1",
      "apiKey": "sk-xxxx-xxxx-xxxx"
    },
    {
      "title": "Opus 4.7 Thinking",
      "provider": "openai",
      "model": "kr/claude-opus-4.7-thinking",
      "apiBase": "https://api.wflab.web.id/v1",
      "apiKey": "sk-xxxx-xxxx-xxxx"
    }
  ]
}

• Provider: OpenAI Compatible
• Base URL: https://api.wflab.web.id/v1
• API Key: sk-xxx
• Model: pilih dari 6 model id di tabel atas

1. Login admin → Settings → Connections
2. + Add OpenAI Connection
3. URL: https://api.wflab.web.id/v1, Key: sk-xxx
4. Save → refresh halaman → 6 model muncul di dropdown chat

Install: pip install openai

from openai import OpenAI

client = OpenAI(
    base_url="https://api.wflab.web.id/v1",
    api_key="sk-xxxx-xxxx-xxxx",
)

resp = client.chat.completions.create(
    model="kr/claude-sonnet-4.6",
    messages=[{"role": "user", "content": "Halo"}],
)
print(resp.choices[0].message.content)

Install: npm install openai

import OpenAI from "openai";

const client = new OpenAI({
  baseURL: "https://api.wflab.web.id/v1",
  apiKey: "sk-xxxx-xxxx-xxxx",
});

const resp = await client.chat.completions.create({
  model: "kr/claude-opus-4.7",
  messages: [{ role: "user", content: "Tulis fungsi fibonacci" }],
});
console.log(resp.choices[0].message.content);

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    base_url="https://api.wflab.web.id/v1",
    api_key="sk-xxxx-xxxx-xxxx",
    model="kr/claude-opus-4.7-thinking",
)

print(llm.invoke("Jelaskan QED").content)

curl https://api.wflab.web.id/v1/chat/completions \
  -H "Authorization: Bearer sk-xxxx-xxxx-xxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kr/claude-sonnet-4.6",
    "messages": [{"role":"user","content":"Halo"}]
  }'

Tambahkan "stream": true untuk streaming SSE.

Hermes adalah agent yang menghubungkan AI ke channel WhatsApp dan Telegram. Untuk pakai API key kami sebagai LLM backend Hermes, isi konfigurasi LLM Provider di Hermes:

Contoh config (file YAML / env)

llm:
  provider: openai
  base_url: https://api.wflab.web.id/v1
  api_key: ${HERMES_LLM_API_KEY}     # set di env
  model: kr/claude-sonnet-4.6
  fallback_model: kr/claude-opus-4.6
  stream: true
  max_tokens: 4096
  timeout: 180

Tips integrasi Hermes

• Pakai sonnet-4.6 sebagai default untuk reply cepat dan hemat credit. Naik ke opus kalau buyer minta reasoning kompleks.
• Aktifkan streaming supaya Hermes bisa kirim chunk ke WA/Telegram secara progresif. Endpoint kami https://api.wflab.web.id/v1/chat/completions support SSE penuh.
• Active IP slot: Hermes biasanya jalan dari 1 server VPS dengan IP tetap, jadi tidak akan kena too_many_active_locations. Kalau Hermes dipindah server, IP lama TTL 5 menit lalu auto-rotate.
• Concurrent limit: kalau Hermes melayani banyak user WA/Telegram bersamaan, pilih plan dengan concurrent yang cukup. Sultan = 4 paralel cukup untuk ~50 user aktif (asumsi 8% concurrent peak).
• Fallback model: set fallback ke model lain dengan weight lebih kecil. Kalau primary kena 502/504 sementara, Hermes auto-retry ke fallback tanpa user merasakan.
• Cek pemakaian rutin di https://api.wflab.web.id/usage untuk memastikan credit Hermes tidak habis di tengah bulan.