Have we quoted something like this before, and at what price?
Describe a new request in plain words and get back your most similar past quotes and what was priced — a reference point, never a recommendation.
- How it works
- Semantic search over your archive of past quotes and deals. A salesperson describes the new request in plain words — 'powder-coated steel brackets, around 500 units, automotive customer' — and gets back the most similar past quotes and what was priced, plus won/lost outcome where you recorded it. Embeddings and the reranker do the matching on the small-model server; the small model writes a two-line summary of each hit.
- Data you need
- Past quotes in machine-readable text form — PDFs with a text layer, CRM records, or a quotes folder. Many small firms have quotes scattered across mailboxes, so a one-off collection effort is the real prerequisite. Won/lost status is a bonus, not a given: most firms never recorded it, and reconstructing it is manual work.
- What to expect
- Similarity is textual, not engineering similarity — two quotes can read alike and be completely different jobs. Old prices are a reference point, never a recommendation: material costs and exchange rates move, and the system has no idea. Sparse archives of a few dozen quotes return weak matches, and scanned or photographed quotes are invisible, since only documents with extractable text can be indexed — no vision or OCR model is served.
- Where people stay involved
- The salesperson judges whether the old quote is genuinely comparable and whether the price still holds. Missing won/lost data means the tool can only show what was quoted, not what worked.
Which model, and what it costs to run
Qwen3-8B
This job answers while someone waits, so latency comes first: we hold it to a single small card and, within that, take the best independent score on sticking to the document.
- 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.
Sticking to the document
Measured on public documents, by a model acting as judge. Read it beside the answer rate: the lowest hallucination rates on this board belong to models that simply decline more often.
Independent measurement · Vectara · board updated May 11, 2026
Reading a long document whole
No benchmark measures this
No current benchmark ranks today's open models here. HELMET showed that the popular test — finding a planted sentence — predicts nothing, and its own table has not been rerun on 2026 models.
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.
- 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.
- 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.
- 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.
Also in sales & b2b
All 15 →Draft a quote from a customer's enquiry email and your price list
Turns an inbound enquiry and your price list into a filled-in quote draft for a salesperson to check and send — a document to correct, not a blank page.
Pull the line items out of RFQ emails and PDFs into a spreadsheet
Extracts part numbers, quantities, units and dates from RFQ emails and PDFs into one consistent table, instead of someone retyping them.
Sort incoming sales emails: real enquiry, existing customer, supplier, junk
Tags every email hitting the sales inbox — new enquiry, existing customer, supplier, junk — within seconds, so real enquiries reach the right person.
Grade new leads against your own definition of a good customer
Scores each new enquiry against your written definition of a good customer, with a short 'why' note the salesperson can read and disagree with.
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.