Introducing prompt_jev(): bringing Jev to Motherduck SQL

MotherDuck shipped prompt_jev(), an integration with TypeSafe AI's Jev model, which classified 100,000 AG News articles in 40 seconds at 89% accuracy for $0.50, versus $37.58 and 31 minutes 59 seconds for gpt-5.6-terra at 88% accuracy. MotherDuck reported Jev ran at 2,484 rows per second, more than 25x better than the compared LLMs across cost, accuracy and speed, and is invoked as a SQL scalar function returning labels, scores or yes/no decisions with confidence. The company said the speed and cost make table-scale text classification viable where LLM prompting or fine-tuning an encoder like BERT was previously impractical.

Introducing prompt jev : bringing Jev to Motherduck SQL - 5 min read Text classification in MotherDuck just got about 50x faster at about 1% of the cost. Today we're shipping an integration with Jev https://typesafe.ai/blog/introducing-system-one-models-and-jev , a new kind of AI model from TypeSafe AI. On a 100,000-row benchmark it matched frontier LLM accuracy in 40 seconds for fifty cents. The comparable LLM took more than half an hour and cost $37. When price gets that low and performance is this good, entire tables that were too expensive to handle are easily within reach. TypeSafe describes the state of AI today as "databases before SQL”, where we first need to understand the database before we can build the universal language on top. Their pitch for Jev is a frontier-intelligence function call: unstructured state goes in, typed probabilistic decisions come out. We read that and thought, well, this fits right into the universal language of SQL. It can be framed as a function that takes a text column and hands back a label, a score, or a yes/no with a confidence score - exactly how scalar functions today. There is no need to parse the response, and so it's immediately ready to filter, join, and aggregate in the same statement. This combination of speed, accuracy, and cost together is what makes new workloads viable. LLMs can already classify text, but running it across a million rows is slow enough and expensive enough it is hard to see the value. The alternative, training an encoder like BERT, means collecting labeled examples and maintaining your own model. This is a big bet for something that may not work until it’s close to prod ready. prompt jev is configured with a few sentences, runs at analytics speed, and is extremely cost effective for the rows it labels. Below, it pulls the main complaint out of a table of support transcripts: SELECT conversation id, prompt jev transcript, 'Identify the customer''s main complaint', choice := {label: 'billing', description: 'Payments, invoices, and refunds'}, {label: 'technical', description: 'Errors, outages, and integrations'}, {label: 'sales', description: 'Pricing and upgrades'}, {label: 'account', description: 'Cancellations and account administration'} AS classification FROM customer conversations; We think this is a very big deal for analytics. A database is, after all, an excellent place to store your million customer conversations. But it’s still really hard to figure out what your customers are complaining about. Before you can track which problems are getting worse, someone or something has to read those conversations and label them. Encoder models like BERT can efficiently classify over a large dataset, but first you have to collect labeled examples, fine-tune the model, and improve the model as new data comes in. This takes patience and expertise. On the other hand, LLMs enable you to just prompt for what you want instead, but running one across a meaningfully large dataset is slow, expensive, and error-prone. prompt jev gives you LLM-style ergonomics with encoder-style efficiency, all at analytics scale. Here are some comparisons that made us excited: | model | rows/s | accuracy | Retail cost/100k | wall time at 100k rows | |---|---|---|---|---| | Jev | 2,484 | 89% | $0.50 | 40s | | gpt-4o-mini | 84 | 80% | $1.93 | 19m 45s | | gpt-5-nano | 94 | 83% | $1.58 | 17m 49s | | gpt-5.6-luna | 61 | 84% | $3.53 | 27m 25s | | gpt-5.6-terra | 52 | 88% | $37.58 | 31m 59s | We benchmarked on 100,000 articles sampled from the training split of AG News https://huggingface.co/datasets/fancyzhx/ag news , the four-class news topic dataset introduced by Zhang, Zhao & LeCun 2015 https://arxiv.org/abs/1509.01626 , scoring each model against the ground truth. What really excited us was that it exceed existing models by 25x across cost, accuracy, and speed dimensions. Furthermore, tests at 1m and 10m rows yielded similar performance and in some cases even faster than our baseline presented above . prompt jev is available on all paid MotherDuck plans. Pick a question you’ve been putting off and try it on your own data Dive into Jev Use our guide to learn how to use MotherDuck and Jev together https://motherduck.com/docs/key-tasks/ai-and-motherduck/classify-text-with-prompt-jev/ for classifying text. Or play around with the Dive below to see what is possible. Further Reading And if you need some inspiration - run the benchmark yourself -- 1 Load AG News train split straight from Hugging Face and draw the 100k sample CREATE TABLE ag train AS SELECT row number OVER AS id, text, label FROM 'hf://datasets/fancyzhx/ag news/data/train-00000-of-00001.parquet'; CREATE TABLE sample 100k AS SELECT FROM ag train USING SAMPLE 100000 ROWS reservoir, 43 ; -- 2 Classify every row with prompt jev CREATE TABLE preds AS SELECT id, label, prompt jev text, 'Classify the topic of this news article.', choice := 'World', 'Sports', 'Business', 'Sci/Tech' AS result FROM sample 100k; -- 3 Score against the dataset labels CREATE MACRO ag name l AS 'World', 'Sports', 'Business', 'Sci/Tech' l + 1 ; -- overall accuracy NULLs reported separately, never scored as wrong SELECT count AS n, count FILTER WHERE result.choice IS NULL AS nulls, round avg result.choice = ag name label ::INT , 4 AS accuracy, round avg result.confidence , 3 AS mean confidence FROM preds; -- per-class precision, recall, F1 WITH c AS SELECT ag name label AS truth, result.choice AS pred FROM preds WHERE result.choice IS NOT NULL , k AS SELECT unnest 'World', 'Sports', 'Business', 'Sci/Tech' AS class SELECT class, count FILTER WHERE truth = class AND pred = class AS tp, count FILTER WHERE truth < class AND pred = class AS fp, count FILTER WHERE truth = class AND pred < class AS fn, round tp / tp + fp , 4 AS "precision", round tp / tp + fn , 4 AS recall, round 2 tp / 2 tp + fp + fn , 4 AS f1 FROM k CROSS JOIN c GROUP BY class ORDER BY class; -- confusion matrix PIVOT SELECT ag name label AS truth, result.choice AS pred FROM preds ON pred USING count GROUP BY truth ORDER BY truth; Table of contents Read Next