ML Env

Use casesSearch & discovery

Match a supplier's price list to your own catalogue, with the machine doing the first pass

Matches a supplier's price list to your product master row by row; confident matches auto-link, everything else queues for a person to decide.

How it works
You get a 4,000-row spreadsheet from a supplier with their product names and codes. For each row, embeddings retrieve the closest candidates from your product master, a reranker tightens the shortlist, and the small language model judges each candidate pair — same product, same variant? Confident matches are linked automatically; everything else queues for a person. This split — embeddings for candidates, an LLM for the pairwise decision — is documented practice in entity matching.
Data you need
Your product master with titles and attributes, and EAN/GTIN codes wherever you have them — a barcode match beats any AI match and short-circuits the pipeline. Plus the supplier file in any tabular form.
What to expect
Expect it to clear the straightforward rows and leave the genuinely ambiguous ones for you; what share is straightforward depends entirely on how your supplier writes product names, so measure it on your own first file rather than trusting any figure. The classic failure mode is variants — pack sizes, colours, 2024-versus-2025 model years look nearly identical to embeddings and often fool the small model too. The model's 'confidence' is not a calibrated probability: set the auto-link threshold on your own data, starting conservative. Heavily abbreviated or foreign-language supplier files degrade retrieval; the embeddings are multilingual, which helps, but truncated codes stay hard.
Where people stay involved
A person reviews every match below the confidence threshold and a sample above it, especially early on. Wrong matches propagate into pricing and stock, so this assists a human rather than replacing one.

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.