Compute

What Compute solves

• Skip provider procurement and contract negotiation — capacity is provisioned on demand.
• Avoid vendor lock-in — the smart scheduler can move workloads to whichever provider has the best cost / latency at the moment of launch.
• Stop guessing your inference bill — every active instance bills a fixed hourly rate.
• Ship faster — the OpenAI contract means your existing tools, SDKs, and evaluation pipelines keep working unchanged.

Capability matrix

OpenAI-compatible contract

Every running instance answers at https://api.qdiv0.com/v1 using the standard OpenAI REST contract. Authentication is a Bearer API key; the model id is the instance serving name. Streaming, function calling, JSON mode, and stop sequences are supported when the underlying runtime exposes them.

1curl https://api.qdiv0.com/v1/chat/completions \
2  -H "Authorization: Bearer $QDIV0_API_KEY" \
3  -H "Content-Type: application/json" \
4  -d '{
5    "model": "<serving-name>",
6    "messages": [{"role": "user", "content": "Hello"}]
7  }'

Pricing model

Compute bills capacity, not tokens. The model below defines how every instance is charged; the Billing page describes the ledger mechanics.

Observability surface

Compute collects a small, opinionated set of signals from every instance. The retention is short for high-cardinality signals and indefinite for things that should never be lost.

The instances list

The Instances page is the day-to-day surface for the models you have deployed. It lists every instance in the account, with state, serving name, and creation time, and exposes the actions you can take on each one.

Per-instance actions

Bulk operations

The checkbox column supports multi-select. Use it to stop or delete many instances at once.

Usage tracking

The Usage tab on each instance tracks two parallel series: cost in EUR and billable seconds of active time. Summary cards at the top of the tab show the month-to-date total, the average daily cost, and the active hours.

Built-in playground

The Playground button on a running instance opens a sheet with three panels: chat, embeddings, and (for embeddings instances) a multimodal input that takes an image URL.

1from openai import OpenAI
2
3client = OpenAI(
4    base_url="https://api.qdiv0.com/v1",
5    api_key="your-api-key",
6)
7
8resp = client.chat.completions.create(
9    model="<serving-name>",
10    messages=[{"role": "user", "content": "Hello"}],
11)
12print(resp.choices[0].message.content)

1import requests
2
3resp = requests.post(
4    "https://api.qdiv0.com/v1/embeddings",
5    headers={"Authorization": f"Bearer {api_key}"},
6    json={"model": "<serving-name>", "input": "hello world"},
7)
8print(resp.json()["data"][0]["embedding"][:8], "...")

1import requests
2
3resp = requests.post(
4    "https://api.qdiv0.com/v1/embeddings",
5    headers={"Authorization": f"Bearer {api_key}"},
6    json={
7        "model": "qwen3-vl-2b-demo",
8        "input": [
9            {"type": "text", "text": "diagram description"},
10            {"type": "image_url", "image_url": {"url": "https://example.com/diagram.png"}},
11        ],
12    },
13)

Schedule rules

The Schedule tab on the instance detail page stores cron-based start and stop rules. Rules are evaluated by the scheduler worker using the rule's own timezone. The last execution (success or error) is shown inline. See Settings for the full cron syntax and supported actions.

Where to go next