Skip to content
Kimi K3 is Arnict's launch model and the $29.99/month Developer Pass is open — Pay As You Go includes $5 of free API usage every month
arnict

Kimi K3 · directly hosted · API coming soon

Private inference. Zero content retained.

Kimi K3 runs on infrastructure Arnict operates. Prompts and completions are returned, then discarded.

KIMI K3PROMPTRESPONSE0 CONTENT RETAINED

Zero data retention

Privacy that procurement can point to.

No badge wall. No vague “we value privacy” copy. The data boundary is stated precisely, published in policy, and made part of the service terms.

Read the Privacy Policy

No prompt or completion storage

Request content exists only while it is needed to return the response. It is not written to application logs, databases, or backups.

No training or advertising use

Customer content is never used to train models, create advertising profiles, or sold. That commitment is written into our Privacy Policy and Terms.

No inference-provider handoff

Kimi K3 runs on Arnict-operated US infrastructure. Your content is not forwarded to a third-party inference API.

Launch model

Kimi K3, with every number in the open.

One launch model, one API identifier, and published token economics. Arnict is positioning for dependable infrastructure—not bargain-bin routing.

API identifier
kimi-k3
Scale
2.8T total · 104B active
Hosting
Arnict-operated · US
Content retention
Zero
USD per 1M tokensLaunch model
Input tokens$2.50
Cached input tokens$0.25
Output tokens$12.50

Pay As You Go

$0/month · $5 of API usage included monthly.

Start free

Developer Pass

$29.99/month · published weekly and monthly allowances.

Compare allowances

Integration

Change the base URL. Keep your client.

Bearer authentication, kimi-k3, and the familiar chat completions shape.

BASE · https://api.arnict.com/v1

POST · /chat/completions

DATA · Zero prompt and completion retention

Open documentation
Illustrative · API coming soon
curl https://api.arnict.com/v1/chat/completions \
  -H "Authorization: Bearer $ARNICT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "kimi-k3",
    "messages": [
      {
        "role": "user",
        "content": "Explain directly hosted inference."
      }
    ]
  }'

A focused launch, with the roadmap kept separate from availability.

Next in priority: Qwen3.8 MAX, then DeepSeek V4 Pro. Every roadmap model remains Upcoming.

View the catalog

Evaluate the architecture before the API opens.

Create an account now, or send procurement and integration questions to one accountable inbox.