ML Env

Use casesSearch & discovery

Reorder your existing search results so the best match comes first

Keep your existing search engine and add a reranking step that re-sorts the top results by relevance before display — the lightest possible upgrade.

How it works
Your shop's existing search engine stays exactly as it is. Its top 50–100 results, plus the query, are passed through a reranking model, which re-sorts them by semantic relevance before display. No index migration, no new search stack — a reranking step bolted onto what you already have.
Data you need
An existing search that returns candidate results with reasonable product text — the reranker can only judge relevance from the text it sees — and a plugin point or API layer where results can be intercepted before rendering. Shopware, Magento and WooCommerce all have one, but wiring it up is a small development job, not a toggle.
What to expect
The cheapest meaningful improvement to a mediocre search: it improves precision, not recall. A reranker can only reorder what your search already found — if the right product never made the candidate list, reranking cannot rescue it; broadening the candidate list is a separate job (semantic search over your catalogue). Quality depends on your product text: thin or duplicated descriptions give the reranker little to work with. It adds a processing step to every query; latency is small in normal operation but worth load-testing at your real traffic levels before a sale event.
Where people stay involved
Merchandising rules — promoted products, margin-based boosts, stock priorities — are applied after reranking, and someone at the shop decides how relevance and commercial boosts trade off. The model only knows relevance, not your margins.

Which model, and what it costs to run

Qwen3-8B

No public benchmark ranks this particular job, so this is not a leaderboard pick — it is where we would start: small, permissively licensed, and strong wherever it has been measured. We prove it on your own content before anything ships.

Licence
Apache-2.0
Weights at 4-bit
5 GB
Context
32K tokens
Publisher
Alibaba (Qwen Team)

What the hardware costs

One 48 GB card holds it

Rent in the EU
$1.60/hrScaleway, Paris (PAR2)
Buy the card
$7,569new, one-off
Or rent it by the token
$0.04 / $0.04per M in / out · DeepInfra

Hardware only, third-party prices from 2026-07. The figure excludes the KV cache, which grows with context length and how many people use it at once — sized properly in a conversation, not guessed here. Renting by the token is cheaper up front; why our customers still self-host is below.

Search over your own catalogue

No benchmark measures this

MTEB and BEIR rank embedding models, but their evaluation sets sit inside the models' training data. Scores fall by double-digit nDCG on a private corpus, so the ranking does not transfer to your catalogue.

So we do not show a chart here. We measure it on your own content, in the first week, and you see the result before anything ships.

The API is cheaper per token. Here is why our customers don't use it.

We will not pretend otherwise: renting a model by the token from a serverless API costs less per million tokens than a card we run for you. We show that price on every use-case page. What it does not include is the part a shop with a customer database actually pays for.

  1. 01

    Your data never leaves hardware you can point at

    A serverless "we don't retain your data" is a clause in a contract. Running the model on a card in Amsterdam is a fact of architecture: your catalogue, tickets and customer records are never sent to a third party at all. For a GDPR audit, that is the difference between a promise and a floor plan.

  2. 02

    The price cannot move without your say-so

    A serverless rate card is somebody else's lever. The provider can raise the price, retire the model, or change the terms, and your cost moves with it. The same model on the same card costs the same next year — you own the number.

  3. 03

    The model cannot be taken away

    Hosted APIs deprecate models on their own schedule; the one you built on can be gone in a quarter. An open-weight model on your own hardware runs for as long as you keep the lights on. No vendor can end-of-life it out from under you.

And the price gap closes with volume: past a card you keep busy — very roughly four billion tokens a month — owning is cheaper outright, even before the three reasons above.

Getting a case like this one from a conversation to production takes about two months, and you can stop at the end of any phase.

Next

Tell us what your team does by hand

Describe the process that takes the most time. We will say plainly whether a model is the right tool for it.