Mary Harrington
Texte de l'e-mail (le modèle a reçu exactement ce texte)
From: Mary Harrington <mary.harrington@hoffman-young.example> To: support@scott-salinas.example Date: Mon, Sep 1, 2025 at 11:12 AM Subject: Re: Order status Hi, I'm following up on the message below, nothing has happened so far. This is completely unacceptable and very frustrating. If this isn't resolved, we will end our business with you. Kind regards, Mary Harrington Hoffman-Young On Fri, Aug 29, 2025 at 2:20 PM, support@scott-salinas.example wrote: > Hi, your request is tracked as ticket 86008, we'll get back to you shortly. > > Scott, Salinas and Mccoy > On Fri, Aug 29, 2025 at 11:12 AM, Mary Harrington <mary.harrington@hoffman-young.example> wrote: > > Hi, > > > > Our order still hasn't arrived, even though it was supposed to be delivered last week. This is about our order 10481757. > > > > Kind regards, > > Mary Harrington > > Hoffman-Young
Verdict
Le moins cher sans erreur : Gemma 4 31B, à 0,12 € pour 1 000 documents.
- Claude Sonnet 5 ne fait aucune erreur non plus, mais coûte 28× plus cher.
- Mistral Small 4 est 1,4× moins cher, mais : valeur inventée (Échéance).
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
| Champ | Valeur correcte | Mistral Small 48 / 9 champs | Gemma 4 31B9 / 9 champs | GPT-6 Luna9 / 9 champs | DeepSeek V4.1 Flash9 / 9 champs | Qwen3.8 Flash9 / 9 champs | Gemini 3.5 Flash Lite9 / 9 champs | Mistral Medium 3.59 / 9 champs | Gemini 3.8 Flash9 / 9 champs | GPT-6 Sol9 / 9 champs | Claude Sonnet 59 / 9 champs |
|---|---|---|---|---|---|---|---|---|---|---|---|
| Catégorie | delivery | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) | delivery (correct) |
| Priorité | high | high (correct) | high (correct) | high (correct) | high (correct) | high (correct) | high (correct) | high (correct) | high (correct) | high (correct) | high (correct) |
| Tonalité | negative | negative (correct) | negative (correct) | negative (correct) | negative (correct) | negative (correct) | negative (correct) | negative (correct) | negative (correct) | negative (correct) | negative (correct) |
| Nom du client | Mary Harrington | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) | Mary Harrington (correct) |
| Numéro de commande | 10481757 | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (correct) | 10481757 (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) |
| Échéance | — | 2025-09-08valeur inventée | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) | — (correct) |
| Tokens | entrée / sortie | 704 / 74 | 748 / 65 | 826 / 100 | 962 / 276 | 1 072 / 491 | 736 / 84 | 704 / 64 | 736 / 478 | 826 / 75 | 1 032 / 159 |
| dont raisonnement caché | facturé comme sortie | 0 | 0 | 46 | 228 | 405 | 0 | 0 | 394 | 21 | 53 |
| dont depuis le cache | entrée facturée à prix réduit | 406 | 453 | 0 | 586 | 768 | 0 | 0 | 0 | 0 | 0 |
| Coût de ce document | cette exécution | 0,0001 € | 0,0001 € | 0,0001 € | 0,0002 € | 0,0003 € | 0,0004 € | 0,0014 € | 0,0022 € | 0,0022 € | 0,0034 € |
| Pour 1 000 documents | pour ce même type de document | 0,09 € | 0,12 € | 0,12 € | 0,25 € | 0,27 € | 0,40 € | 1,41 € | 2,16 € | 2,21 € | 3,36 € |
| Latence | cette exécution | 0,9 s | 3,5 s | 2,3 s | 3,7 s | 6,9 s | 0,7 s | 0,8 s | 1,9 s | 2,7 s | 3,3 s |