GLMAvailable
Long-context chat, coding and tool workflows with GLM 5.3 Flash.
GLM 5.3 Flash
1,048,576 context
Open-model inference
Open models.
Private by
design.
Build with open models through one OpenAI-compatible API. Clear token pricing. Your prompts stay yours.
Create an account. Get a key. Make your first call. Quickstart →
- Private by designNo prompt or response archive
- Affordable pricingClear rates. Pay for what you use.
- Fast throughputBuilt for responsive applications
About ArnictBuild your app.
Build your app.
Choose your model.
Add open-model inference to the tools you already use. Compare models, stream responses, and track what each request costs.
A familiar way to build.
Use the OpenAI SDK with your Arnict key, our base URL, and an available model.
Make your first call →Your prompts stay yours.
Inference content is never archived or used for training. See how we handle requests and usage data.
Read about privacy →
Integrate
Create an account, get an API key, and choose an available model. Use the OpenAI SDK to send your first request.
- Stream text as the model responds
- Create and revoke API keys in your dashboard
- Review request status, token usage and cost
Keep your client. Change the endpoint.
Node.js
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api.arnict.com/v1",
apiKey: process.env.ARNICT_API_KEY,
});
const completion = await client.chat.completions.create({
model: process.env.ARNICT_MODEL, // an available model ID from the catalog
messages: [{ role: "user", content: "Summarize this ticket." }],
stream: true,
});
for await (const chunk of completion) {
process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}Questions
Yes. Set the base URL to https://api.arnict.com/v1 and use an Arnict API key. Arnict supports chat completions and streaming. Check the model page for supported capabilities and limits.
Inference content is processed to answer your request. It is never archived as conversation history or used for training or advertising. Temporary processing caches are discarded through expiry, eviction or the end of the process that holds them. Account and usage records include metadata such as model, tokens, status and cost.
Serverless inference is available for models marked Available in the catalog. LoRA, full fine-tuning (FFT), and dedicated GPU inference for a custom model of your choice are coming soon.
Each model has separate rates for input, cached input and output tokens, quoted per million tokens. Current rates appear in the catalog. Review per-request usage and cost in your dashboard.
Yes. Planned support includes LoRA and full fine-tuning, plus dedicated inference for customer-chosen custom models. These features are coming soon. Training and dedicated inference guides explain the scope and how to follow updates.
Share Arnict. Earn for life.
15% on a referral’s first deposit, then 10% on later deposits, in usage credits. Earnings confirm as purchased credits are used; only unused credit waits 14 days. Above $250 in referred paid compute spend, negotiate cash terms.
Get started
Choose an open model, connect your app, and get your first response.
From an idea to your first response.
