Verdicts / Structured JSON extraction from messy documents
mistralai/mistral-small-3.2-24b-instruct is the cheapest model that matches openai/gpt-4.1
Structured JSON extraction from messy documents: 150 public examples, 8 models, retries and failed attempts counted in every price. Checked 2026-11-04.
JSON extraction · 150 public examples
mistralai/mistral-small-3.2-24b-instruct
Matches openai/gpt-4.1 on Structured JSON extraction from messy documents, at a fraction of the price.
Cost per successful task = total OpenRouter usage.cost across every attempt (retries on invalid output or API errors included) ÷ tasks that passed. Attempts per task above 1.00 is the retry tax. Status: Cleared = pass rate and the lower end of its 95% interval are at or above the 55.0% bar; Borderline = pass rate at the bar, interval reaches below it; Below bar = pass rate under it.
Why it counts as a match
Pass rates on the same tasks can be compared pair by pair. mistralai/mistral-small-3.2-24b-instruct failed 3 tasks that openai/gpt-4.1 passed, and passed 4 that the baseline failed, across 150 shared tasks. An exact one-sided test of "no more than 5 points worse" gives p = 0.000000, so it is shown to match within the margin. Pass rates: 62.7% vs 62.0%.
What the switch pays back
Your monthly bill on the baseline
Monthly saving at this rate
Days to repay a $19 run
Days to repay a $49 run
$200
$192.15
3
7.7
$500
$480.38
1.2
3.1
$2,000
$1,921.52
0.3
0.8
Saving = bill × (1 − $0.000203 ÷ $0.005173), the cost per successful task measured on this dataset. Your own bill and examples set your own number; a Clearbar run on your data replaces this estimate with a measured one.
Where mistralai/mistral-small-3.2-24b-instruct fails
94 of 150 tasks passed. These are the first 12 that did not, with the input, what was expected and what the model returned.
**SOFTWARE LICENSE AGREEMENT** This Software License Agreement (the "Agreement") is entered into as of the 8th day of March, 2023, by and between: 1. Acme Corporation, a corporation organized and existing under the laws of the State of California, with its principal place of business
Expected
{"agreementDate":"2023-03-08","parties":[{"name":"Acme Corporation","address":"123 Main Street, Anytown, CA 12345","email":"info@acmecorp.com"},{"name":"XYZ Software, Inc.","address":"456 Elm Street, Anytown, CA 98765","email":"sales@xyzsoftware.com"}],"softwareProduct":{"name":"XYZ Software Suite", …
Got
located at 123 Business Ave, Suite 100, San Francisco, CA 94107, USA, and 2. Global Tech Solutions, a corporation organized and existing under the laws of the United Kingdom, with its principal place of business located at 10 Downing St, London, SW1A 2AA, UK. **RECITALS** WHEREAS, Acme Corporation ( …
Why it failed
wrong fields: license
**Business Report** **Date:** 2023-03-08 **Company:** Acme Corporation **Report Period:** January 1, 2023 - March 31, 2023 **Executive Summary** Acme Corporation experienced a strong financial performance in the first quarter of 2023. Revenue increased by 15% compared to the previous quarter, and expenses remained rela …
Expected
{"businessName":"Acme Corporation","reportDate":"2023-03-08","revenue":100000,"expenses":[{"category":"Salaries and Wages","amount":50000},{"category":"Rent and Utilities","amount":10000},{"category":"Marketing and Advertising","amount":5000}],"profit":35000,"notes":"This report shows a strong finan …
Got
```json { "businessName": "Acme Corporation", "reportDate": "2023-03-08", "revenue": 100000, "expenses": [ { "category": "Total expenses", "amount": 65000 } ], "profit": 35000, "notes": "Acme Corporation is optimistic about its financial outlook for the remainder of 2023. The company expects to cont …
Why it failed
wrong fields: expenses, notes
## Technical Documentation: Device Configuration ### Introduction This document provides a comprehensive overview of the configuration details for a specific device, identified as 'device-01'. The device is categorized as a sensor, specifically a Temperature Sensor, and is employed to measure the temperature within a l …
Expected
{"deviceId":"device-01","deviceType":"sensor","deviceName":"Temperature Sensor","deviceDescription":"Measures the temperature in the living room.","dataCollectionInterval":600,"sensorConfigs":[{"sensorId":"sensor-01","sensorType":"temperature","reportingInterval":10}]}
Got
```json { "deviceId": "device-01", "deviceType": "sensor", "deviceName": "Temperature Sensor", "deviceDescription": "This sensor is responsible for measuring the temperature in the living room.", "dataCollectionInterval": 600, "sensorConfigs": { "sensorId": "sensor-01", "sensorType": "temperature", …
Why it failed
wrong fields: deviceDescription, sensorConfigs
## Technical Documentation: Assembly of Product X ### Introduction This document provides detailed instructions for assembling Product X. It includes information on the required materials, tools, safety requirements, and estimated time for each step of the assembly process. ### Work Instruction ID: WI-12345 ### Name: A …
Expected
{"id":"WI-12345","name":"Assembly of Product X","description":"This work instruction describes the steps required to assemble Product X.","version":"1.0","steps":[{"id":"S-1","name":"Gather materials","description":"Gather all the materials required for assembly.","resources":["Component A","Compone …
Got
```json { "id": "WI-12345", "name": "Assembly of Product X", "description": "This work instruction describes the steps required to assemble Product X. It is intended for use by trained personnel who are familiar with the assembly process and the associated safety requirements.", "version": "1.0", "s …
Why it failed
wrong fields: description, steps
Subject: Feedback on Product ID: product_456 from User ID: user_123 Dear [Recipient Name], I hope this email finds you well. I am writing to provide you with valuable feedback from one of our esteemed customers, user_123, regarding product_456. On March 8, 2023, user_123 made a purchase of product_456 and has since pro …
Expected
{"userId":"user_123","productId":"product_456","rating":4,"comment":"This product is great! It's easy to use and has a lot of features.","purchaseDate":"2023-03-08","satisfaction":true,"recommendations":["Add more colors to the product.","Make the product more affordable."]}
Got
```json { "userId": "user_123", "productId": "product_456", "rating": 4, "comment": "user_123 enthusiastically praised the product's ease of use and the abundance of features it offers. They specifically mentioned that the product has made their daily tasks more efficient and enjoyable.", "purchaseD …
Why it failed
wrong fields: comment, recommendations
# Data Classification Policy ## Overview This document defines the data classification policy for the organization. The policy specifies the rules for classifying data based on its sensitivity and the actions that should be taken to protect it. ## Policy Name The name of the data classification policy is "My Data Class …
Expected
{"name":"My Data Classification Policy","description":"This policy defines the data classification rules for my organization.","rules":[{"condition":{"field":"project","operator":"EQUALS","value":"my-project"},"classification":"CONFIDENTIAL"},{"condition":{"field":"dataset","operator":"STARTS_WITH", …
Got
```json { "name": "My Data Classification Policy", "description": "The purpose of this policy is to ensure that all data within the organization is classified and protected appropriately. The policy defines the data classification rules and the actions that should be taken to protect data based on i …
Why it failed
wrong fields: description
**Sales Receipt** **Transaction ID:** cr-123456 **Date:** 2023-03-08 **Customer ID:** cu-987654 **Store ID:** st-246810 **Offer ID:** of-654321 **Coupon Code:** ABC123 **Amount Paid:** 10.00 **Status:** PROCESSED **Additional Information:** * The customer purchased an item using the specified coupon code. * The item wa …
wrong fields: id, offerId, customerId, storeId, couponCode, amountPaid, status
## Guitar Maintenance Guide This document provides a comprehensive guide to maintaining your guitar, ensuring its optimal performance and longevity. ### Instrument Type: Guitar The guitar is a versatile and popular instrument, known for its rich sound and diverse musical applications. It consists of a body, neck, strin …
Expected
{"instrumentType":"Guitar","maintenanceTasks":[{"taskName":"Clean the guitar strings","taskDescription":"Use a soft cloth to wipe away any dirt or grime from the strings.","taskFrequency":"Weekly"},{"taskName":"Tune the guitar","taskDescription":"Use a tuner to ensure that the guitar is in tune.","t …
Got
```json { "instrumentType": "Guitar", "maintenanceTasks": [ { "taskName": "Clean the Guitar Strings", "taskDescription": "Use a soft, dry cloth to gently wipe away any dirt, grime, or fingerprints from the strings. Avoid using harsh chemicals or abrasive materials, as these can damage the strings.", …
Why it failed
wrong fields: maintenanceTasks, additionalInfo
## Technical Documentation: "The Hitchhiker's Guide to the Galaxy" ### Introduction: "The Hitchhiker's Guide to the Galaxy" is a renowned science fiction comedy novel written by the esteemed author Douglas Adams. Initially published in 1979 by Pan Books, this captivating tale has garnered widespread acclaim for its wit …
Expected
{"title":"The Hitchhiker's Guide to the Galaxy","authors":[{"firstName":"Douglas","lastName":"Adams"}],"publisher":"Pan Books","publicationYear":1979,"identifiers":[{"type":"ISBN","value":"0330258648"}],"pageCount":184,"price":{"amount":12.99,"currency":"USD"},"summary":"A hilarious and thought-prov …
Subject: My First Todo List - A Comprehensive Update Dear [Recipient Name], I hope this email finds you well. I wanted to share an update on my first todo list, which I've been diligently working on since its creation. The list is titled "My First Todo List" and serves as a comprehensive guide to my tasks and goals. It …
Expected
{"id":"1","name":"My First Todo List","description":"This is my first todo list.","tasks":[{"id":"1","title":"Learn JSON","description":"I need to learn JSON for my new job.","priority":"high","due_date":"2023-02-28","completed":false},{"id":"2","title":"Build a website","description":"I want to bui …
Got
```json { "TodoList": { "id": "My First Todo List", "name": "My First Todo List", "description": "A comprehensive guide to my tasks and goals.", "tasks": [ { "id": "Learn JSON", "title": "Learn JSON", "description": "This task is crucial for my new job, and I've set a high priority for it. The due d …
Why it failed
wrong fields: id, name, tasks
**Social Media Influencer Jane Doe Shares Insights on the Importance of Social Media Marketing** In a recent post on Twitter, social media influencer Jane Doe (@janedoe) shared her latest blog post on the importance of social media marketing. The post, which has garnered over 100 likes, 20 comments, and 50 shares, high …
Expected
{"platform":"Twitter","timestamp":"2023-03-08T18:30:00Z","content":{"text":"Check out my latest blog post on the importance of social media marketing! #socialmediamarketing #digitalmarketing","image":"https://example.com/image.jpg","link":"https://example.com/blog/social-media-marketing"},"likes":10 …
Got
```json { "platform": "Twitter", "timestamp": "2023-10-05T12:00:00Z", "content": { "text": "Social media marketing is no longer just a trend; it's a necessity for businesses of all sizes. With billions of people using social media platforms every day, it's an incredibly powerful tool for reaching an …
Why it failed
wrong fields: content, author
## Technical Documentation: Product Brochure ### Overview: The Product Brochure is a marketing material designed to showcase a company's products and services. It serves as an informative and visually appealing tool to engage potential customers and provide detailed information about the company's offerings. ### Dimens …
Expected
{"name":"Product Brochure","description":"A brochure showcasing our company's products and services.","dimensions":{"height":210,"width":297},"material":"paper","ink":"water-based","colors":["black","white","blue"],"fonts":["Arial","Helvetica","Times New Roman"],"images":[{"url":"https://example.com …
Got
```json { "name": "Product Brochure", "description": "A marketing material designed to showcase a company's products and services. It serves as an informative and visually appealing tool to engage potential customers and provide detailed information about the company's offerings.", "dimensions": { " …
Why it failed
wrong fields: description, texts
Where the baseline openai/gpt-4.1 fails
**SOFTWARE LICENSE AGREEMENT** This Software License Agreement (the "Agreement") is entered into as of the 8th day of March, 2023, by and between: 1. Acme Corporation, a corporation organized and existing under the laws of the State of California, with its principal place of business
Expected
{"agreementDate":"2023-03-08","parties":[{"name":"Acme Corporation","address":"123 Main Street, Anytown, CA 12345","email":"info@acmecorp.com"},{"name":"XYZ Software, Inc.","address":"456 Elm Street, Anytown, CA 98765","email":"sales@xyzsoftware.com"}],"softwareProduct":{"name":"XYZ Software Suite", …
**Hotel Receptionist:** Good morning, Ms. Doe. Welcome to the Grandview Hotel. How may I assist you today? **Jane Doe:** I'm here to check in for my reservation. **Hotel Receptionist:** Certainly, Ms. Doe. May I have your hotel and room number, please? **Jane Doe:** It's HTL-1234 and RM-5678. **Hotel Receptionist:** Th …
Expected
{"hotelId":"HTL-1234","roomId":"RM-5678","guest":{"firstName":"Jane","lastName":"Doe","email":"jane.doe@example.com"},"reservation":{"startDate":"2023-08-15","endDate":"2023-08-19","numNights":4,"numGuests":2},"specialRequests":"Please provide a room with a view of the pool."}
Got
```json { "hotelId": "HTL-1234", "roomId": "1012", "guest": { "firstName": "Jane", "lastName": "Doe", "email": null }, "reservation": { "startDate": "2023-08-15", "endDate": "2023-08-19", "numNights": 4, "numGuests": 2 }, "specialRequests": "Room with a view of the pool" } ```
Why it failed
wrong fields: roomId
**Business Report** **Date:** 2023-03-08 **Company:** Acme Corporation **Report Period:** January 1, 2023 - March 31, 2023 **Executive Summary** Acme Corporation experienced a strong financial performance in the first quarter of 2023. Revenue increased by 15% compared to the previous quarter, and expenses remained rela …
Expected
{"businessName":"Acme Corporation","reportDate":"2023-03-08","revenue":100000,"expenses":[{"category":"Salaries and Wages","amount":50000},{"category":"Rent and Utilities","amount":10000},{"category":"Marketing and Advertising","amount":5000}],"profit":35000,"notes":"This report shows a strong finan …
Got
{ "businessName": "Acme Corporation", "reportDate": "2023-03-08", "revenue": 100000, "expenses": [], "profit": 35000, "notes": "Revenue increased by 15% compared to the previous quarter, totaling $100,000. Expenses for the first quarter were $65,000, a decrease of $2,000 (3%) compared to the previou …
Why it failed
wrong fields: expenses, notes
**Breaking News: Jane Doe Receives a New Message from John Smith** In a recent development, Jane Doe, a prominent figure in the tech industry, has received a new message from John Smith, a renowned business magnate. The message, sent on March 8, 2023, at 6:30 PM UTC, holds significant importance due to the professional …
Expected
{"id":"629b609b-471e-4073-855f-54347882941d","type":"user","recipient":{"id":"e462322e-571b-493b-9334-301a1d108196","name":"Jane Doe","email":"jane.doe@example.com"},"message":{"id":"20613336-6324-487c-a138-540739591124","title":"New message from John Smith","body":"Hi Jane, I hope you're having a g …
Got
{ "id": null, "type": "user", "recipient": { "id": null, "name": "Jane Doe", "email": null }, "message": { "id": null, "title": "New message from John Smith", "body": "The message contains a cordial greeting and expresses John Smith's hope that Jane Doe is having a wonderful day. The body of the mes …
Why it failed
wrong fields: message
## Technical Documentation: Device Configuration ### Introduction This document provides a comprehensive overview of the configuration details for a specific device, identified as 'device-01'. The device is categorized as a sensor, specifically a Temperature Sensor, and is employed to measure the temperature within a l …
Expected
{"deviceId":"device-01","deviceType":"sensor","deviceName":"Temperature Sensor","deviceDescription":"Measures the temperature in the living room.","dataCollectionInterval":600,"sensorConfigs":[{"sensorId":"sensor-01","sensorType":"temperature","reportingInterval":10}]}
Got
{ "deviceId": "device-01", "deviceType": "sensor", "deviceName": "Temperature Sensor", "deviceDescription": "This sensor is responsible for measuring the temperature in the living room.", "dataCollectionInterval": 600, "actuatorConfigs": null, "sensorConfigs": { "sensorId": "sensor-01", "sensorType" …
Why it failed
wrong fields: deviceDescription, sensorConfigs
## Technical Documentation: Assembly of Product X ### Introduction This document provides detailed instructions for assembling Product X. It includes information on the required materials, tools, safety requirements, and estimated time for each step of the assembly process. ### Work Instruction ID: WI-12345 ### Name: A …
Expected
{"id":"WI-12345","name":"Assembly of Product X","description":"This work instruction describes the steps required to assemble Product X.","version":"1.0","steps":[{"id":"S-1","name":"Gather materials","description":"Gather all the materials required for assembly.","resources":["Component A","Compone …
Got
{ "id": "WI-12345", "name": "Assembly of Product X", "description": "This work instruction describes the steps required to assemble Product X. It is intended for use by trained personnel who are familiar with the assembly process and the associated safety requirements.", "version": "1.0", "steps": [ …
Why it failed
wrong fields: description, steps
Subject: Feedback on Product ID: product_456 from User ID: user_123 Dear [Recipient Name], I hope this email finds you well. I am writing to provide you with valuable feedback from one of our esteemed customers, user_123, regarding product_456. On March 8, 2023, user_123 made a purchase of product_456 and has since pro …
Expected
{"userId":"user_123","productId":"product_456","rating":4,"comment":"This product is great! It's easy to use and has a lot of features.","purchaseDate":"2023-03-08","satisfaction":true,"recommendations":["Add more colors to the product.","Make the product more affordable."]}
Got
{ "userId": "user_123", "productId": "product_456", "rating": 4, "comment": "Praised the product's ease of use and abundance of features, stating it has made daily tasks more efficient and enjoyable.", "purchaseDate": "2023-03-08", "satisfaction": true, "recommendations": [ "Expand the product's col …
Why it failed
wrong fields: comment, recommendations
# Data Classification Policy ## Overview This document defines the data classification policy for the organization. The policy specifies the rules for classifying data based on its sensitivity and the actions that should be taken to protect it. ## Policy Name The name of the data classification policy is "My Data Class …
Expected
{"name":"My Data Classification Policy","description":"This policy defines the data classification rules for my organization.","rules":[{"condition":{"field":"project","operator":"EQUALS","value":"my-project"},"classification":"CONFIDENTIAL"},{"condition":{"field":"dataset","operator":"STARTS_WITH", …
Got
{ "name": "My Data Classification Policy", "description": "The purpose of this policy is to ensure that all data within the organization is classified and protected appropriately. The policy defines the data classification rules and the actions that should be taken to protect data based on its class …
Why it failed
wrong fields: description
**MEDICAL DOCUMENT** **Patient:** Jane Doe **Date:** 2023-08-15 **Assessment Reason:** Initial assessment **Patient Information:** * Name: Jane Doe * Date of Birth: 1980-01-01 * Gender: Female **Mental Health Professional:** * Name: Dr. Smith * Credentials: MD, PhD **Diagnostic Impression:** * Depression, moderate **Tr …
Expected
{"assessmentReason":"Initial assessment","patient":{"firstName":"Jane","lastName":"Doe","dateOfBirth":"1980-01-01","gender":"Female"},"mentalHealthProfessional":{"name":"Dr. Smith","credentials":"MD, PhD"},"diagnosticImpression":"Depression, moderate","treatmentPlan":"Cognitive-behavioral therapy an …
## Guitar Maintenance Guide This document provides a comprehensive guide to maintaining your guitar, ensuring its optimal performance and longevity. ### Instrument Type: Guitar The guitar is a versatile and popular instrument, known for its rich sound and diverse musical applications. It consists of a body, neck, strin …
Expected
{"instrumentType":"Guitar","maintenanceTasks":[{"taskName":"Clean the guitar strings","taskDescription":"Use a soft cloth to wipe away any dirt or grime from the strings.","taskFrequency":"Weekly"},{"taskName":"Tune the guitar","taskDescription":"Use a tuner to ensure that the guitar is in tune.","t …
Got
{ "instrumentType": "Guitar", "maintenanceTasks": [ { "taskName": "Clean the Guitar Strings", "taskDescription": "Use a soft, dry cloth to gently wipe away any dirt, grime, or fingerprints from the strings. Avoid using harsh chemicals or abrasive materials, as these can damage the strings.", "taskFr …
Why it failed
wrong fields: maintenanceTasks, additionalInfo
## Technical Documentation: "The Hitchhiker's Guide to the Galaxy" ### Introduction: "The Hitchhiker's Guide to the Galaxy" is a renowned science fiction comedy novel written by the esteemed author Douglas Adams. Initially published in 1979 by Pan Books, this captivating tale has garnered widespread acclaim for its wit …
Expected
{"title":"The Hitchhiker's Guide to the Galaxy","authors":[{"firstName":"Douglas","lastName":"Adams"}],"publisher":"Pan Books","publicationYear":1979,"identifiers":[{"type":"ISBN","value":"0330258648"}],"pageCount":184,"price":{"amount":12.99,"currency":"USD"},"summary":"A hilarious and thought-prov …
Subject: My First Todo List - A Comprehensive Update Dear [Recipient Name], I hope this email finds you well. I wanted to share an update on my first todo list, which I've been diligently working on since its creation. The list is titled "My First Todo List" and serves as a comprehensive guide to my tasks and goals. It …
Expected
{"id":"1","name":"My First Todo List","description":"This is my first todo list.","tasks":[{"id":"1","title":"Learn JSON","description":"I need to learn JSON for my new job.","priority":"high","due_date":"2023-02-28","completed":false},{"id":"2","title":"Build a website","description":"I want to bui …
Got
{ "id": null, "name": "My First Todo List", "description": "A comprehensive guide to my tasks and goals.", "tasks": [ { "id": "1", "title": "Learn JSON", "description": "This task is crucial for my new job, and I've set a high priority for it. I'm actively engaged in learning the intricacies of JSON …
Temperature 0, max tokens 1500, up to 2 retries on invalid output or API errors. System prompt:
Extract the data described by the JSON schema from the text. Reply with one JSON object that follows the schema, and nothing else. Fill every field the text states; use null only for a field the text does not mention.
Re-run it yourself
The script runs this exact dataset and model list through OpenRouter with your own key. Models sample at temperature 0 and are not fully deterministic, so a re-run lands within about ±5 points of pass rate and ±10% of cost. Prices move: today's list may differ from the 2026-11-04 snapshot.
Prices and models move every month. Leave your email and we'll send a note when a verdict here changes. One short email per change. Or follow the feed.