ML Env

Use casesVoice of customer

Code the open answers from your post-purchase survey overnight

Codes the open answers from your post-purchase surveys into countable categories overnight, with a short readable digest per question.

How it works
'How did you hear about us?', 'What nearly stopped you buying?', 'Anything we should improve?' — the answers pile up unread because coding them by hand is tedious. The small model assigns each answer to categories from a codebook, one codebook per question, and the large model writes a short narrative summary per question with representative quotes. The large model can also propose a first-draft codebook from a sample of answers, which a human then edits and approves.
Data you need
Survey exports with each open question's answers as a column — any survey tool exports this — plus a codebook per question, often drafted by the model from a sample of a couple of hundred answers and then edited by a human.
What to expect
Pays off at volume: with only a trickle of answers each month, reading them by hand is quicker than maintaining a codebook. One-word answers and 'n/a' clutter need filtering rules or the counts get distorted; answers touching several categories get single-labelled unless you explicitly allow multi-label, which makes the counting harder to read. Comparing waves over time only works if the codebook is frozen between waves — every edit breaks the trend line — and mixed-language answer sets need a per-language plan agreed up front, since that is where quality drifts.
Where people stay involved
A human approves the codebook before the bulk run and spot-checks a sample of coded answers. The model handles the volume; a person owns the category definitions and the edge cases. Expect some miscoding on ambiguous, sarcastic or off-topic answers — the spot check is what tells you whether that rate is tolerable.

Which model, and what it costs to run

Qwen3-8B

This job runs in bulk rather than to a waiting person, so size is not the constraint — we take the strongest independent score on sticking to the document that we may serve freely and that fits on a single card.

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

Vectara Hallucination Leaderboard · HHEM · 17 of 18 models measured

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.

Phi-43.7% · answers 80.7% ⚠Llama 3.3 70B4.1% · answers 99.5%Gemma 3 12B4.4% · answers 97.4%Qwen3-8B4.8% · answers 99.9%Mistral Small 3.25.1% · answers 97.9%Granite 4.0 Small5.2% · answers 100%DeepSeek-V3.25.3% · answers 96.6%Qwen3-14B5.4% · answers 99.9%
Shorter is betterFree to serveConditions apply⚠ answered under 95%

Independent measurement · Vectara · board updated May 11, 2026

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.

Also in voice of customer

All 15 →

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.