The right model for each request

Smart routing picks the model and provider for each request. Send "auto" and DigitSpace chooses the cheapest model that can handle what the request needs at the quality level you set; name a model and it picks the best provider for that model.

Start free Read the docs

Asked forauto (balanced)

Needstools, 2,100 tokens in

Shortlist, per 1M tokens in / out

claude-haiku-5-5Anthropic · $0.10 / $0.50
gpt-6-lunaOpenAI · $0.10 / $0.50
gemini-3.8-flashGoogle · $0.75 / $3.75

Chosen: claude-haiku-5-5 on Anthropic, first attempt

Why one model for everything costs too much

Most requests don't need the biggest model. Summarising a ticket, tagging a message or rewriting a sentence is done just as well by a model that costs a tenth as much.

Hard-coding one model means paying the top price for every request, or getting weak answers on the hard ones. Routing decides per request instead.

Telephone switchboard operators with headsets at a long switchboard, plugging cords into the panel. — In 1953 an operator put each call through by hand. DigitSpace routes every request.
Switchboard operators, NIH Clinical Center, 1953. NIH History Office, public domain, via Wikimedia Commons.

The three quality levels

Set a level for the project, or ask for one in the request with "auto:economy", "auto:balanced" or "auto:best".

Level	Models allowed	Good for
Economy	Every model in the auto pool	High volume, simple text: tagging, summaries, extraction
Balanced (default)	Standard and advanced models	Most product features and chat
Best	Advanced models only	Hard reasoning, long agent runs, code

Requests that need tools or reasoning always get at least a standard model, whatever the level.

How a model is chosen

Needs.
The request is read for images, tools, a JSON schema, reasoning and its size.
Shortlist.
Only models that support all of it, and that your rules allow, stay on the list.
Rank.
Healthy providers first, then by your strategy: price, latency, or a 50/50 balance.
Answer.
The response says which model answered, and the "x-digitspace-provider" header names the provider.

Use it

Change "model" to "auto". Nothing else in your code changes.

from openai import OpenAI

client = OpenAI(base_url="https://api.digitspace.ai/v1", api_key=DIGITSPACE_API_KEY)

reply = client.chat.completions.create(
    model="auto:balanced",  # or "auto", "auto:economy", "auto:best"
    messages=[{"role": "user", "content": "Summarise this ticket in two sentences."}],
)

Keep control when you need it

Pin a model: Name it, e.g. "claude-sonnet-5-5". Routing then only chooses among its providers.
Pin a provider: Name "anthropic/claude-sonnet-5-5" to always use that provider.
Choose the strategy: Send "X-DigitSpace-Routing: price" or "latency" on a request, or set it for the project.

Questions

A telephone exchange room around 1900, with a long wooden switchboard, operator stools and a desk. — General Telephone Company of the Northwest, exchange room, about 1900, public domain, cropped, via Wikimedia Commons.

Which models can auto choose?

The models in the auto pool, which our team keeps current and labels basic, standard or advanced. Your project's rules can narrow it further.

Will auto change models without telling me?

Each response names the model that answered, and every request is logged with it, so you can always see what was used.

Does auto cost extra?

No. You pay the chosen model's price like any other request.

Can I use auto with the Anthropic API?

Yes. Send "auto" as the model to /v1/messages and the answer comes back in Anthropic's format, whichever model served it.

Live in a minute.

Create a key, change the base URL, send a request. No card needed with your own provider keys.

Start free