AI agents
Tool calling and long context for agents that take actions.
AI inference for builders.
Dedicated inference. One OpenAI-compatible API. Pay as you go, no subscription.
- Tokens processed
- 165B+
- Fast inference: tokens per second, per request
- Up to 350
- Zero data retention
- ZDR
Works with pi, OpenCode, Cline, n8n, the OpenAI SDK and any tool that supports the OpenAI API.
Build with it
Automation workflows
Connect n8n, Python scripts or any OpenAI-compatible tool.
RAG and document search
Embeddings and chat models in one account.
Your own AI product
Build on the API and bill your users your way.
Models
DeepSeek-V4.1-Flash-1017
Newby DeepSeek · 1M context · Original precision
- Coding
- Agents
- Reasoning
- Long context
- Input
- US$0.22
US$0.30 -
- Output
- US$0.90
US$1.20 -
per 1M tokens
Qwen3.8-27B
Newby Qwen · 256K context · FP8
- Coding
- Agents
- Reasoning
- Input
- US$0.02
US$0.04 -
- Output
- US$1.25
US$2.50 -
per 1M tokens
GLM-5.3-Flash
Newby Z.ai · 1M context · NVFP4
- Coding
- Agents
- Writing
- Input
- US$0.12
- Output
- US$0.40
per 1M tokens
MiniMax-H3
Newby MiniMax · Video with audio, up to 768p · FP8
- Text to video
- Image to video
- Standard
- US$0.05
- Turbo
- US$0.025
per second of video
Qwen3-Embedding-8B
Newby Qwen · 32K context · BF16
- Search
- RAG
- Classification
- Input
- US$0.01
per 1M tokens
Usage and billing
Clear pricing
Per-token rates, published up front.
Usage reports
Tokens and cost per model, team and key.
Team access
Add members and set limits per key.
Prepaid or postpaid
Top up in the console, with optional auto top-up.
Usage overview
- Total tokens
- 2,482,991
- Estimated cost
- Rp 248.299
Dedicated hardware
Models run on our own dedicated servers.
Hosted in Indonesia
Servers and data stay in Indonesia.
Your data stays yours
We don't store prompts or outputs, and we don't train on them.
How it works
Sign up
Create an account. Add your team later.
Add credit
Top up in the console. Turn on auto top-up if you like.
Get your API key
Create a key and start sending requests. Works with the OpenAI SDK.
Pricing
| Model | Input per 1M | Output per 1M | Cached input per 1M | Context |
|---|---|---|---|---|
DeepSeek-V4.1-Flash-1017by DeepSeek · Original precision |
US$0.22 |
US$0.90 |
US$0.0042 |
1M |
Qwen3.8-27Bby Qwen · FP8 |
US$0.02 |
US$1.25 |
US$0.01 |
256K |
GLM-5.3-Flashby Z.ai · NVFP4 |
US$0.12 | US$0.40 | US$0.022 | 1M |
MiniMax-H3by MiniMax · FP8 |
per second of video |
768p | ||
Qwen3-Embedding-8Bby Qwen · BF16 |
US$0.01 | 32K | ||
US$ per 1M tokens or per second of video. You pay the rate in effect when the request runs.
Rupiah prices follow the current exchange rate and may change.
Cached input is text the model already processed in a recent request, like long instructions you send every time.
FAQ
Which apps work with Fenrir?
Anything that supports the OpenAI API, including pi, OpenCode, Cline, n8n and the OpenAI SDK. Point it at the Fenrir base URL and use your Fenrir key.
How am I billed?
Per token, at the model's rate when each request runs. Prepaid accounts use their balance. Postpaid accounts get a monthly statement.
Do you offer zero data retention (ZDR)?
Yes. We don't store your prompts or outputs, and we don't train on them. We keep only request details like time, model, token counts, cost and status, for billing and support.
Can my team use one account?
Yes. The account owner adds members and sets limits per key.
Can I cancel anytime?
Yes. There is no subscription. You only pay for what you use.
How do I contact you?
Email us at [email protected] or message us on Instagram at @lio.fenrir.