ModelVerdict

Russell Bates

Anglais (US)E-mail9 champs

Document en-p0004 · mise en page plain · 9 champs · partie publique du jeu de données · Même facture · Tchèque → · Signaler une erreur dans cet exemple

Texte de l'e-mail (le modèle a reçu exactement ce texte)

From: Russell Bates <russell.bates@reed-walker.example>
To: support@mendoza.example
Date: Mon, Mar 16, 2026 at 1:55 PM
Subject: Can't log in

Hi,

The data export in your app fails with an error and the file is empty. We need an answer by April 9.

Thanks,
Russell Bates
Reed-Walker

Verdict

Le moins cher sans erreur : Gemma 4 31B, à 0,10 € pour 1 000 documents.

  • Claude Sonnet 5 ne fait aucune erreur non plus, mais coûte 42,7× plus cher.
  • Mistral Small 4 est 1,6× moins cher, mais : valeur erronée (Priorité).

Prompt soumis à chaque modèle

You triage the support inbox. Read the email below and return only JSON with these keys:

category        one of: billing (invoices, payments, charges), delivery (where is the order, shipping,
                address change), complaint (damaged, wrong or defective goods, warranty claims),
                technical (login, app or system errors, integrations), cancellation (cancel an order,
                contract or subscription), sales (quotes, prices, interest before buying)
priority        one of: urgent, high, normal, low (rules below)
sentiment       one of: positive, neutral, negative (the customer's tone)
customer_name   full name of the customer who wrote the request
order_number    the customer's order number, if mentioned
invoice_number  invoice number, if mentioned
amount          the amount of money the request is about (number)
currency        ISO 4217 code of that amount (USD, EUR, CZK, …)
deadline        the date by which the customer wants a resolution or answer (YYYY-MM-DD)

Priority rules:
- urgent: the customer cannot work right now, or the deadline is at most 1 day after the email date
- high: the deadline is at most 7 days after the email date, or the customer threatens to cancel
  or escalate (lawyer, ending the business)
- low: the customer says explicitly that it is not urgent
- normal: everything else

Rules:
- A thread: judge the newest message; order numbers, invoice numbers and amounts may come from the
  quoted earlier messages. A forwarded email: the customer is the original sender, not the
  colleague who forwarded it.
- Resolve relative deadlines ("by Friday", "by tomorrow", "by the end of the month") against the
  date of the message that states them.
- Support ticket numbers, tax IDs and phone numbers are not order or invoice numbers.
- If a value is not in the email, return null. Do not guess.
- Amounts as numbers with a decimal point, no thousands separators and no currency symbol.

Notre notation

Correspondance exacte par champ, après normalisation des dates, des montants et des espaces. Les diacritiques manquants comptent comme une erreur. Le coût correspond à la consommation réelle de tokens de cette exécution, convertie au taux indiqué en bas de page.

Ce que chaque modèle a extrait

ChampValeur correcteMistral Small 47 / 9 champsGemma 4 31B9 / 9 champsGPT-6 Luna9 / 9 champsDeepSeek V4.1 Flash8 / 9 champsQwen3.8 Flash9 / 9 champsGemini 3.5 Flash Lite8 / 9 champsMistral Medium 3.58 / 9 champsGPT-6 Sol9 / 9 champsGemini 3.8 Flash9 / 9 champsClaude Sonnet 59 / 9 champs
Catégorietechnicaltechnical (correct)technical (correct)technical (correct)technical (correct)technical (correct)technical (correct)technical (correct)technical (correct)technical (correct)technical (correct)
Prioriténormal highvaleur erronéenormal (correct)normal (correct)normal (correct)normal (correct)normal (correct) highvaleur erronéenormal (correct)normal (correct)normal (correct)
Tonaliténeutral negativevaleur erronéeneutral (correct)neutral (correct) negativevaleur erronéeneutral (correct) negativevaleur erronéeneutral (correct)neutral (correct)neutral (correct)neutral (correct)
Nom du clientRussell BatesRussell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)Russell Bates (correct)
Numéro de commande—— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)
Numéro de facture—— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)
Montant—— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)
Devise—— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)— (correct)
Échéance2026-04-092026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)2026-04-09 (correct)
Tokensentrée / sortie530 / 68556 / 67661 / 108802 / 185878 / 695544 / 87530 / 68661 / 82544 / 632774 / 292
dont raisonnement cachéfacturé comme sortie00511356060025545187
dont depuis le cacheentrée facturée à prix réduit4064058676800000
Coût de ce documentcette exécution0,0001 €0,0001 €0,0001 €0,0002 €0,0003 €0,0004 €0,0012 €0,0020 €0,0026 €0,0041 €
Pour 1 000 documentspour ce même type de document0,06 €0,10 €0,11 €0,16 €0,33 €0,35 €1,20 €1,97 €2,56 €4,11 €
Latencecette exécution0,8 s1,9 s2,5 s2,6 s10,4 s0,8 s1,2 s2,7 s3,2 s5,6 s