Claude Haiku 4.5

import { createClient } from '@runware/sdk'

const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()

const [result] = await client.run({
  model: 'anthropic:claude@haiku-4.5',
  settings: {
    systemPrompt: 'You are a precise operations assistant. When given an image of notes, extract only what is visible, distinguish uncertain readings, and present the result in a practical structured summary. Do not invent text that cannot be reasonably read.',
    maxTokens: 1200,
    thinkingLevel: 'low',
    cache: {
      scope: 'system',
      ttl: '5m'
    }
  },
  inputs: {
    images: [
      'https://assets.runware.ai/assets/inputs/5449a50d-44d8-4804-ae32-1d4ec14aef52.jpg'
    ]
  },
  messages: [
    {
      role: 'user',
      content: 'Analyze this planning-board photo and return: 1) a one-paragraph overview of the event, 2) a chronological schedule in bullet points, 3) a table-like list of tasks with owners if visible, 4) supplies or budget concerns mentioned, and 5) any text that is unclear or partially illegible.'
    }
  ]
})

import asyncio
import os

from runware import Runware


async def main():
    async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
        results = await client.run({
            "model": "anthropic:claude@haiku-4.5",
            "settings": {
                "systemPrompt": "You are a precise operations assistant. When given an image of notes, extract only what is visible, distinguish uncertain readings, and present the result in a practical structured summary. Do not invent text that cannot be reasonably read.",
                "maxTokens": 1200,
                "thinkingLevel": "low",
                "cache": {
                    "scope": "system",
                    "ttl": "5m"
                }
            },
            "inputs": {
                "images": [
                    "https://assets.runware.ai/assets/inputs/5449a50d-44d8-4804-ae32-1d4ec14aef52.jpg"
                ]
            },
            "messages": [
                {
                    "role": "user",
                    "content": "Analyze this planning-board photo and return: 1) a one-paragraph overview of the event, 2) a chronological schedule in bullet points, 3) a table-like list of tasks with owners if visible, 4) supplies or budget concerns mentioned, and 5) any text that is unclear or partially illegible."
                }
            ]
        })


asyncio.run(main())

curl https://api.runware.ai/v1 \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "taskType": "textInference",
      "taskUUID": "cccb39e2-8aaf-43b7-b587-825a563eb703",
      "model": "anthropic:claude@haiku-4.5",
      "settings": {
        "systemPrompt": "You are a precise operations assistant. When given an image of notes, extract only what is visible, distinguish uncertain readings, and present the result in a practical structured summary. Do not invent text that cannot be reasonably read.",
        "maxTokens": 1200,
        "thinkingLevel": "low",
        "cache": {
          "scope": "system",
          "ttl": "5m"
        }
      },
      "inputs": {
        "images": [
          "https://assets.runware.ai/assets/inputs/5449a50d-44d8-4804-ae32-1d4ec14aef52.jpg"
        ]
      },
      "messages": [
        {
          "role": "user",
          "content": "Analyze this planning-board photo and return: 1) a one-paragraph overview of the event, 2) a chronological schedule in bullet points, 3) a table-like list of tasks with owners if visible, 4) supplies or budget concerns mentioned, and 5) any text that is unclear or partially illegible."
        }
      ]
    }
  ]'

runware run anthropic:claude@haiku-4.5 \
  settings.systemPrompt="You are a precise operations assistant. When given an image of notes, extract only what is visible, distinguish uncertain readings, and present the result in a practical structured summary. Do not invent text that cannot be reasonably read." \
  settings.maxTokens=1200 \
  settings.thinkingLevel=low \
  settings.cache.scope=system \
  settings.cache.ttl=5m \
  inputs.images.0=https://assets.runware.ai/assets/inputs/5449a50d-44d8-4804-ae32-1d4ec14aef52.jpg \
  messages.0.role=user \
  messages.0.content="Analyze this planning-board photo and return: 1) a one-paragraph overview of the event, 2) a chronological schedule in bullet points, 3) a table-like list of tasks with owners if visible, 4) supplies or budget concerns mentioned, and 5) any text that is unclear or partially illegible."

{
  "taskType": "textInference",
  "taskUUID": "cccb39e2-8aaf-43b7-b587-825a563eb703",
  "model": "anthropic:claude@haiku-4.5",
  "settings": {
    "systemPrompt": "You are a precise operations assistant. When given an image of notes, extract only what is visible, distinguish uncertain readings, and present the result in a practical structured summary. Do not invent text that cannot be reasonably read.",
    "maxTokens": 1200,
    "thinkingLevel": "low",
    "cache": {
      "scope": "system",
      "ttl": "5m"
    }
  },
  "inputs": {
    "images": [
      "https://assets.runware.ai/assets/inputs/5449a50d-44d8-4804-ae32-1d4ec14aef52.jpg"
    ]
  },
  "messages": [
    {
      "role": "user",
      "content": "Analyze this planning-board photo and return: 1) a one-paragraph overview of the event, 2) a chronological schedule in bullet points, 3) a table-like list of tasks with owners if visible, 4) supplies or budget concerns mentioned, and 5) any text that is unclear or partially illegible."
    }
  ]
}

import { createClient } from '@runware/sdk'

const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()

const [result] = await client.run({
  model: 'anthropic:claude@haiku-4.5',
  settings: {
    maxTokens: 1200,
    thinkingLevel: 'medium'
  },
  messages: [
    {
      role: 'user',
      content: 'Turn the raw notes below into a leadership-ready incident brief. Reconcile the chronology, distinguish confirmed facts from estimates, and do not invent missing details.\n\nRAW NOTES\n- Incident: checkout API elevated failures in EU-West\n- Deployment began 09:08 UTC\n- Alerts fired 09:12\n- Checkout failures above baseline from 09:12 to 09:37\n- Peak failure rate: 18.6% of checkout attempts at 09:19\n- Other regions unaffected\n- On-call identified database connection exhaustion at 09:17\n- Rollback initiated 09:24 and completed 09:31\n- Error rate returned to baseline at 09:37\n- No evidence of duplicate charges, lost orders, or data corruption\n- Analytics estimates 4.2% of affected shoppers abandoned checkout, but this is not directly measured\n- Support received 41 related tickets\n- Preliminary cause: release changed the checkout service connection-pool limit from 120 to 30; the lower limit became insufficient during a traffic surge\n- Release passed staging because staging load tests did not reproduce production concurrency\n- Immediate mitigation: rollback\n- Follow-ups: Priya to add a connection-pool regression test by 18 September; Mateo to add a production-concurrency scenario to load testing by 25 September; Lin to audit connection limits across payment services by 20 September\n- Still unknown: exact number of lost purchases and why the configuration review did not flag the change\n\nOUTPUT FORMAT\n1. Status — one line\n2. Executive summary — maximum 80 words\n3. Customer and business impact — bullets, labeling estimates explicitly\n4. Timeline — compact UTC table\n5. Cause and contributing factors\n6. Corrective actions — table with owner and due date\n7. Open questions\n8. Customer-facing update — maximum 60 words, plain language, no internal implementation details\n\nUse a calm, accountable, blameless tone. Keep the complete brief under 450 words.'
    }
  ]
})

import asyncio
import os

from runware import Runware


async def main():
    async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
        results = await client.run({
            "model": "anthropic:claude@haiku-4.5",
            "settings": {
                "maxTokens": 1200,
                "thinkingLevel": "medium"
            },
            "messages": [
                {
                    "role": "user",
                    "content": "Turn the raw notes below into a leadership-ready incident brief. Reconcile the chronology, distinguish confirmed facts from estimates, and do not invent missing details.\n\nRAW NOTES\n- Incident: checkout API elevated failures in EU-West\n- Deployment began 09:08 UTC\n- Alerts fired 09:12\n- Checkout failures above baseline from 09:12 to 09:37\n- Peak failure rate: 18.6% of checkout attempts at 09:19\n- Other regions unaffected\n- On-call identified database connection exhaustion at 09:17\n- Rollback initiated 09:24 and completed 09:31\n- Error rate returned to baseline at 09:37\n- No evidence of duplicate charges, lost orders, or data corruption\n- Analytics estimates 4.2% of affected shoppers abandoned checkout, but this is not directly measured\n- Support received 41 related tickets\n- Preliminary cause: release changed the checkout service connection-pool limit from 120 to 30; the lower limit became insufficient during a traffic surge\n- Release passed staging because staging load tests did not reproduce production concurrency\n- Immediate mitigation: rollback\n- Follow-ups: Priya to add a connection-pool regression test by 18 September; Mateo to add a production-concurrency scenario to load testing by 25 September; Lin to audit connection limits across payment services by 20 September\n- Still unknown: exact number of lost purchases and why the configuration review did not flag the change\n\nOUTPUT FORMAT\n1. Status — one line\n2. Executive summary — maximum 80 words\n3. Customer and business impact — bullets, labeling estimates explicitly\n4. Timeline — compact UTC table\n5. Cause and contributing factors\n6. Corrective actions — table with owner and due date\n7. Open questions\n8. Customer-facing update — maximum 60 words, plain language, no internal implementation details\n\nUse a calm, accountable, blameless tone. Keep the complete brief under 450 words."
                }
            ]
        })


asyncio.run(main())

curl https://api.runware.ai/v1 \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "taskType": "textInference",
      "taskUUID": "65d6f55b-cefd-4c6b-abef-9a44bbb7d715",
      "model": "anthropic:claude@haiku-4.5",
      "settings": {
        "maxTokens": 1200,
        "thinkingLevel": "medium"
      },
      "messages": [
        {
          "role": "user",
          "content": "Turn the raw notes below into a leadership-ready incident brief. Reconcile the chronology, distinguish confirmed facts from estimates, and do not invent missing details.\n\nRAW NOTES\n- Incident: checkout API elevated failures in EU-West\n- Deployment began 09:08 UTC\n- Alerts fired 09:12\n- Checkout failures above baseline from 09:12 to 09:37\n- Peak failure rate: 18.6% of checkout attempts at 09:19\n- Other regions unaffected\n- On-call identified database connection exhaustion at 09:17\n- Rollback initiated 09:24 and completed 09:31\n- Error rate returned to baseline at 09:37\n- No evidence of duplicate charges, lost orders, or data corruption\n- Analytics estimates 4.2% of affected shoppers abandoned checkout, but this is not directly measured\n- Support received 41 related tickets\n- Preliminary cause: release changed the checkout service connection-pool limit from 120 to 30; the lower limit became insufficient during a traffic surge\n- Release passed staging because staging load tests did not reproduce production concurrency\n- Immediate mitigation: rollback\n- Follow-ups: Priya to add a connection-pool regression test by 18 September; Mateo to add a production-concurrency scenario to load testing by 25 September; Lin to audit connection limits across payment services by 20 September\n- Still unknown: exact number of lost purchases and why the configuration review did not flag the change\n\nOUTPUT FORMAT\n1. Status — one line\n2. Executive summary — maximum 80 words\n3. Customer and business impact — bullets, labeling estimates explicitly\n4. Timeline — compact UTC table\n5. Cause and contributing factors\n6. Corrective actions — table with owner and due date\n7. Open questions\n8. Customer-facing update — maximum 60 words, plain language, no internal implementation details\n\nUse a calm, accountable, blameless tone. Keep the complete brief under 450 words."
        }
      ]
    }
  ]'

runware run anthropic:claude@haiku-4.5 \
  settings.maxTokens=1200 \
  settings.thinkingLevel=medium \
  messages.0.role=user \
  messages.0.content="Turn the raw notes below into a leadership-ready incident brief. Reconcile the chronology, distinguish confirmed facts from estimates, and do not invent missing details.

RAW NOTES
- Incident: checkout API elevated failures in EU-West
- Deployment began 09:08 UTC
- Alerts fired 09:12
- Checkout failures above baseline from 09:12 to 09:37
- Peak failure rate: 18.6% of checkout attempts at 09:19
- Other regions unaffected
- On-call identified database connection exhaustion at 09:17
- Rollback initiated 09:24 and completed 09:31
- Error rate returned to baseline at 09:37
- No evidence of duplicate charges, lost orders, or data corruption
- Analytics estimates 4.2% of affected shoppers abandoned checkout, but this is not directly measured
- Support received 41 related tickets
- Preliminary cause: release changed the checkout service connection-pool limit from 120 to 30; the lower limit became insufficient during a traffic surge
- Release passed staging because staging load tests did not reproduce production concurrency
- Immediate mitigation: rollback
- Follow-ups: Priya to add a connection-pool regression test by 18 September; Mateo to add a production-concurrency scenario to load testing by 25 September; Lin to audit connection limits across payment services by 20 September
- Still unknown: exact number of lost purchases and why the configuration review did not flag the change

OUTPUT FORMAT
1. Status — one line
2. Executive summary — maximum 80 words
3. Customer and business impact — bullets, labeling estimates explicitly
4. Timeline — compact UTC table
5. Cause and contributing factors
6. Corrective actions — table with owner and due date
7. Open questions
8. Customer-facing update — maximum 60 words, plain language, no internal implementation details

Use a calm, accountable, blameless tone. Keep the complete brief under 450 words."

{
  "taskType": "textInference",
  "taskUUID": "65d6f55b-cefd-4c6b-abef-9a44bbb7d715",
  "model": "anthropic:claude@haiku-4.5",
  "settings": {
    "maxTokens": 1200,
    "thinkingLevel": "medium"
  },
  "messages": [
    {
      "role": "user",
      "content": "Turn the raw notes below into a leadership-ready incident brief. Reconcile the chronology, distinguish confirmed facts from estimates, and do not invent missing details.\n\nRAW NOTES\n- Incident: checkout API elevated failures in EU-West\n- Deployment began 09:08 UTC\n- Alerts fired 09:12\n- Checkout failures above baseline from 09:12 to 09:37\n- Peak failure rate: 18.6% of checkout attempts at 09:19\n- Other regions unaffected\n- On-call identified database connection exhaustion at 09:17\n- Rollback initiated 09:24 and completed 09:31\n- Error rate returned to baseline at 09:37\n- No evidence of duplicate charges, lost orders, or data corruption\n- Analytics estimates 4.2% of affected shoppers abandoned checkout, but this is not directly measured\n- Support received 41 related tickets\n- Preliminary cause: release changed the checkout service connection-pool limit from 120 to 30; the lower limit became insufficient during a traffic surge\n- Release passed staging because staging load tests did not reproduce production concurrency\n- Immediate mitigation: rollback\n- Follow-ups: Priya to add a connection-pool regression test by 18 September; Mateo to add a production-concurrency scenario to load testing by 25 September; Lin to audit connection limits across payment services by 20 September\n- Still unknown: exact number of lost purchases and why the configuration review did not flag the change\n\nOUTPUT FORMAT\n1. Status — one line\n2. Executive summary — maximum 80 words\n3. Customer and business impact — bullets, labeling estimates explicitly\n4. Timeline — compact UTC table\n5. Cause and contributing factors\n6. Corrective actions — table with owner and due date\n7. Open questions\n8. Customer-facing update — maximum 60 words, plain language, no internal implementation details\n\nUse a calm, accountable, blameless tone. Keep the complete brief under 450 words."
    }
  ]
}

import { createClient } from '@runware/sdk'

const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()

const [result] = await client.run({
  model: 'anthropic:claude@haiku-4.5',
  settings: {
    systemPrompt: 'You are a meticulous archival transcription assistant. Extract information conservatively, clearly separate what is directly visible from what is inferred, and preserve ambiguity when handwriting is unclear. Prefer concise structured output.',
    maxTokens: 1400,
    thinkingLevel: 'low',
    cache: {
      scope: 'system',
      ttl: '1h'
    }
  },
  inputs: {
    images: [
      'https://assets.runware.ai/assets/inputs/0d7e480f-ca3a-4d06-a5ab-38c5dd55e8db.jpg'
    ]
  },
  messages: [
    {
      role: 'user',
      content: 'Analyze the uploaded ledger page and return:\n1. A 2-3 sentence overview of what kind of document this is.\n2. A markdown table with one row per visible entry and these columns: line_number, accession_id, date, donor, object_description, dimensions, condition_notes, confidence.\n3. A bullet list of any illegible words or uncertain readings.\n4. A short preservation note describing stains, folds, fading, tears, or annotations visible on the page.\nIf a field is missing, use "—". Do not invent text that is not reasonably readable.'
    }
  ]
})

import asyncio
import os

from runware import Runware


async def main():
    async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
        results = await client.run({
            "model": "anthropic:claude@haiku-4.5",
            "settings": {
                "systemPrompt": "You are a meticulous archival transcription assistant. Extract information conservatively, clearly separate what is directly visible from what is inferred, and preserve ambiguity when handwriting is unclear. Prefer concise structured output.",
                "maxTokens": 1400,
                "thinkingLevel": "low",
                "cache": {
                    "scope": "system",
                    "ttl": "1h"
                }
            },
            "inputs": {
                "images": [
                    "https://assets.runware.ai/assets/inputs/0d7e480f-ca3a-4d06-a5ab-38c5dd55e8db.jpg"
                ]
            },
            "messages": [
                {
                    "role": "user",
                    "content": "Analyze the uploaded ledger page and return:\n1. A 2-3 sentence overview of what kind of document this is.\n2. A markdown table with one row per visible entry and these columns: line_number, accession_id, date, donor, object_description, dimensions, condition_notes, confidence.\n3. A bullet list of any illegible words or uncertain readings.\n4. A short preservation note describing stains, folds, fading, tears, or annotations visible on the page.\nIf a field is missing, use \"—\". Do not invent text that is not reasonably readable."
                }
            ]
        })


asyncio.run(main())

curl https://api.runware.ai/v1 \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "taskType": "textInference",
      "taskUUID": "4ff75869-4fb1-4889-a0d1-1c1d03f4d03c",
      "model": "anthropic:claude@haiku-4.5",
      "settings": {
        "systemPrompt": "You are a meticulous archival transcription assistant. Extract information conservatively, clearly separate what is directly visible from what is inferred, and preserve ambiguity when handwriting is unclear. Prefer concise structured output.",
        "maxTokens": 1400,
        "thinkingLevel": "low",
        "cache": {
          "scope": "system",
          "ttl": "1h"
        }
      },
      "inputs": {
        "images": [
          "https://assets.runware.ai/assets/inputs/0d7e480f-ca3a-4d06-a5ab-38c5dd55e8db.jpg"
        ]
      },
      "messages": [
        {
          "role": "user",
          "content": "Analyze the uploaded ledger page and return:\n1. A 2-3 sentence overview of what kind of document this is.\n2. A markdown table with one row per visible entry and these columns: line_number, accession_id, date, donor, object_description, dimensions, condition_notes, confidence.\n3. A bullet list of any illegible words or uncertain readings.\n4. A short preservation note describing stains, folds, fading, tears, or annotations visible on the page.\nIf a field is missing, use \"—\". Do not invent text that is not reasonably readable."
        }
      ]
    }
  ]'

runware run anthropic:claude@haiku-4.5 \
  settings.systemPrompt="You are a meticulous archival transcription assistant. Extract information conservatively, clearly separate what is directly visible from what is inferred, and preserve ambiguity when handwriting is unclear. Prefer concise structured output." \
  settings.maxTokens=1400 \
  settings.thinkingLevel=low \
  settings.cache.scope=system \
  settings.cache.ttl=1h \
  inputs.images.0=https://assets.runware.ai/assets/inputs/0d7e480f-ca3a-4d06-a5ab-38c5dd55e8db.jpg \
  messages.0.role=user \
  messages.0.content="Analyze the uploaded ledger page and return:
1. A 2-3 sentence overview of what kind of document this is.
2. A markdown table with one row per visible entry and these columns: line_number, accession_id, date, donor, object_description, dimensions, condition_notes, confidence.
3. A bullet list of any illegible words or uncertain readings.
4. A short preservation note describing stains, folds, fading, tears, or annotations visible on the page.
If a field is missing, use \"—\". Do not invent text that is not reasonably readable."

{
  "taskType": "textInference",
  "taskUUID": "4ff75869-4fb1-4889-a0d1-1c1d03f4d03c",
  "model": "anthropic:claude@haiku-4.5",
  "settings": {
    "systemPrompt": "You are a meticulous archival transcription assistant. Extract information conservatively, clearly separate what is directly visible from what is inferred, and preserve ambiguity when handwriting is unclear. Prefer concise structured output.",
    "maxTokens": 1400,
    "thinkingLevel": "low",
    "cache": {
      "scope": "system",
      "ttl": "1h"
    }
  },
  "inputs": {
    "images": [
      "https://assets.runware.ai/assets/inputs/0d7e480f-ca3a-4d06-a5ab-38c5dd55e8db.jpg"
    ]
  },
  "messages": [
    {
      "role": "user",
      "content": "Analyze the uploaded ledger page and return:\n1. A 2-3 sentence overview of what kind of document this is.\n2. A markdown table with one row per visible entry and these columns: line_number, accession_id, date, donor, object_description, dimensions, condition_notes, confidence.\n3. A bullet list of any illegible words or uncertain readings.\n4. A short preservation note describing stains, folds, fading, tears, or annotations visible on the page.\nIf a field is missing, use \"—\". Do not invent text that is not reasonably readable."
    }
  ]
}

import { createClient } from '@runware/sdk'

const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()

const [result] = await client.run({
  model: 'anthropic:claude@haiku-4.5',
  settings: {
    systemPrompt: 'You are a precise document-understanding assistant. Read shopping receipts carefully, preserve uncertainty when text is ambiguous, and present the result in clean JSON followed by a short plain-language note.',
    maxTokens: 1200,
    thinkingLevel: 'low',
    cache: {
      scope: 'system',
      ttl: '1h'
    }
  },
  inputs: {
    images: [
      'https://assets.runware.ai/assets/inputs/c4adf6ec-1fd5-4f40-9393-40e5ae30f839.jpg'
    ]
  },
  messages: [
    {
      role: 'user',
      content: 'Analyze the receipt image and return: 1) merchant_name, 2) purchase_datetime, 3) currency if inferable, 4) an items array with description, quantity if shown, unit_price if shown, line_total, 5) subtotal, tax, total, 6) payment_method details if visible, 7) a short list of possible OCR ambiguities, and 8) whether the arithmetic appears consistent. First output valid JSON only, then on a new line write a concise one-sentence summary.'
    }
  ]
})

import asyncio
import os

from runware import Runware


async def main():
    async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
        results = await client.run({
            "model": "anthropic:claude@haiku-4.5",
            "settings": {
                "systemPrompt": "You are a precise document-understanding assistant. Read shopping receipts carefully, preserve uncertainty when text is ambiguous, and present the result in clean JSON followed by a short plain-language note.",
                "maxTokens": 1200,
                "thinkingLevel": "low",
                "cache": {
                    "scope": "system",
                    "ttl": "1h"
                }
            },
            "inputs": {
                "images": [
                    "https://assets.runware.ai/assets/inputs/c4adf6ec-1fd5-4f40-9393-40e5ae30f839.jpg"
                ]
            },
            "messages": [
                {
                    "role": "user",
                    "content": "Analyze the receipt image and return: 1) merchant_name, 2) purchase_datetime, 3) currency if inferable, 4) an items array with description, quantity if shown, unit_price if shown, line_total, 5) subtotal, tax, total, 6) payment_method details if visible, 7) a short list of possible OCR ambiguities, and 8) whether the arithmetic appears consistent. First output valid JSON only, then on a new line write a concise one-sentence summary."
                }
            ]
        })


asyncio.run(main())

curl https://api.runware.ai/v1 \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "taskType": "textInference",
      "taskUUID": "c8fe03e6-46ef-47b7-8a0c-ee97dd7fa060",
      "model": "anthropic:claude@haiku-4.5",
      "settings": {
        "systemPrompt": "You are a precise document-understanding assistant. Read shopping receipts carefully, preserve uncertainty when text is ambiguous, and present the result in clean JSON followed by a short plain-language note.",
        "maxTokens": 1200,
        "thinkingLevel": "low",
        "cache": {
          "scope": "system",
          "ttl": "1h"
        }
      },
      "inputs": {
        "images": [
          "https://assets.runware.ai/assets/inputs/c4adf6ec-1fd5-4f40-9393-40e5ae30f839.jpg"
        ]
      },
      "messages": [
        {
          "role": "user",
          "content": "Analyze the receipt image and return: 1) merchant_name, 2) purchase_datetime, 3) currency if inferable, 4) an items array with description, quantity if shown, unit_price if shown, line_total, 5) subtotal, tax, total, 6) payment_method details if visible, 7) a short list of possible OCR ambiguities, and 8) whether the arithmetic appears consistent. First output valid JSON only, then on a new line write a concise one-sentence summary."
        }
      ]
    }
  ]'

runware run anthropic:claude@haiku-4.5 \
  settings.systemPrompt="You are a precise document-understanding assistant. Read shopping receipts carefully, preserve uncertainty when text is ambiguous, and present the result in clean JSON followed by a short plain-language note." \
  settings.maxTokens=1200 \
  settings.thinkingLevel=low \
  settings.cache.scope=system \
  settings.cache.ttl=1h \
  inputs.images.0=https://assets.runware.ai/assets/inputs/c4adf6ec-1fd5-4f40-9393-40e5ae30f839.jpg \
  messages.0.role=user \
  messages.0.content="Analyze the receipt image and return: 1) merchant_name, 2) purchase_datetime, 3) currency if inferable, 4) an items array with description, quantity if shown, unit_price if shown, line_total, 5) subtotal, tax, total, 6) payment_method details if visible, 7) a short list of possible OCR ambiguities, and 8) whether the arithmetic appears consistent. First output valid JSON only, then on a new line write a concise one-sentence summary."

{
  "taskType": "textInference",
  "taskUUID": "c8fe03e6-46ef-47b7-8a0c-ee97dd7fa060",
  "model": "anthropic:claude@haiku-4.5",
  "settings": {
    "systemPrompt": "You are a precise document-understanding assistant. Read shopping receipts carefully, preserve uncertainty when text is ambiguous, and present the result in clean JSON followed by a short plain-language note.",
    "maxTokens": 1200,
    "thinkingLevel": "low",
    "cache": {
      "scope": "system",
      "ttl": "1h"
    }
  },
  "inputs": {
    "images": [
      "https://assets.runware.ai/assets/inputs/c4adf6ec-1fd5-4f40-9393-40e5ae30f839.jpg"
    ]
  },
  "messages": [
    {
      "role": "user",
      "content": "Analyze the receipt image and return: 1) merchant_name, 2) purchase_datetime, 3) currency if inferable, 4) an items array with description, quantity if shown, unit_price if shown, line_total, 5) subtotal, tax, total, 6) payment_method details if visible, 7) a short list of possible OCR ambiguities, and 8) whether the arithmetic appears consistent. First output valid JSON only, then on a new line write a concise one-sentence summary."
    }
  ]
}

import { createClient } from '@runware/sdk'

const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()

const [result] = await client.run({
  model: 'anthropic:claude@haiku-4.5',
  settings: {
    systemPrompt: 'You are a meticulous operations analyst. Resolve inconsistencies, preserve important details, and present findings in crisp sections with practical recommendations.',
    maxTokens: 8192,
    thinkingLevel: 'medium',
    cache: {
      scope: 'system',
      ttl: '1h'
    }
  },
  messages: [
    {
      role: 'user',
      content: 'You are helping a small historical-instrument workshop prepare for a guild inspection. Below is a messy internal note compiled from voice memos, sticky notes, and partial shift logs. Read it carefully, then do four things: (1) produce a clean inventory table with item name, expected count, confirmed count, status, and notes; (2) list every contradiction or ambiguity you found; (3) identify the top 5 issues that must be resolved before inspection, ranked by urgency; (4) draft a short message to the workshop lead summarizing what to fix today. Be conservative: if something is uncertain, say so explicitly.\n\nWORKSHOP NOTE START\n\nNorth bench cabinet:\n- Brass astrolabe blanks: should be 18 after Tuesday casting.\n- Mira said 2 were set aside for engraving practice, but I only saw one tagged practice blank.\n- Finished rete assemblies on cork board: 7 total? Wait, no — board has 6 hanging, plus 1 wrapped in linen in drawer N3.\n- Drawer N3 label says \'rete assembly, polished x2\', but one bundle felt too light.\n\nLathe corner:\n- Ebony handles for demonstration pieces: 12 planned for the season.\n- Counted 9 in the tray at open.\n- Tomas borrowed 2 yesterday for photography props; not sure if returned.\n- Afternoon note says \'all handles back except cracked one\' — unclear whether cracked one is included in the tray count or discarded.\n\nApprentice shelf:\n- Star charts, printed reference sheets: master set should be 24 sheets.\n- Current stack clipped in blue folder: 23 sheets.\n- One coffee-stained copy was removed last week, but Hana may have reprinted it.\n- There is also a loose duplicate of sheet 7 with handwritten angle corrections.\n\nFinishing room:\n- Lacquer jars, clear coat: purchase ledger says 8 jars delivered.\n- Visible on rack: 5 sealed, 1 open, 1 nearly empty.\n- Shipping crate by door still contains packing straw and maybe one jar, but nobody checked because the lid is jammed.\n- Safety log says minimum reserve before inspection is 6 usable jars.\n\nFront display prep:\n- Demonstration astrolabes ready for visitors: target 5.\n- Window stand currently holds 3.\n- One more complete unit on polishing cloth near the vice.\n- Another unit is missing its pointer pin; body and dial are assembled.\n\nTool roll inventory:\n- Fine gravers expected: 14.\n- Roll pockets occupied: 13.\n- One graver may be at sharpening station.\n- Sharpening station note says \'returned after honing\', but timestamp smudged and no initials.\n\nPacking area:\n- Wool wrap squares for transport kits: expected 30.\n- Open bundle counted at 26.\n- Two kits already pre-packed with 2 wraps each.\n- Scrap bin contains 1 wrap cut into strips, not usable.\n\nGeneral remarks:\n- If an item is assembled but missing a single essential fastening piece, do not count it as inspection-ready.\n- Practice blanks should not count toward finished demonstration inventory.\n- Consumables that are open but still usable may count as usable reserve, but nearly empty containers should be flagged separately.\n- If two notes conflict, prefer the most recent note only when it is specific.\n\nWORKSHOP NOTE END\n\nReturn the answer in plain text with clear headings.'
    }
  ]
})

import asyncio
import os

from runware import Runware


async def main():
    async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
        results = await client.run({
            "model": "anthropic:claude@haiku-4.5",
            "settings": {
                "systemPrompt": "You are a meticulous operations analyst. Resolve inconsistencies, preserve important details, and present findings in crisp sections with practical recommendations.",
                "maxTokens": 8192,
                "thinkingLevel": "medium",
                "cache": {
                    "scope": "system",
                    "ttl": "1h"
                }
            },
            "messages": [
                {
                    "role": "user",
                    "content": "You are helping a small historical-instrument workshop prepare for a guild inspection. Below is a messy internal note compiled from voice memos, sticky notes, and partial shift logs. Read it carefully, then do four things: (1) produce a clean inventory table with item name, expected count, confirmed count, status, and notes; (2) list every contradiction or ambiguity you found; (3) identify the top 5 issues that must be resolved before inspection, ranked by urgency; (4) draft a short message to the workshop lead summarizing what to fix today. Be conservative: if something is uncertain, say so explicitly.\n\nWORKSHOP NOTE START\n\nNorth bench cabinet:\n- Brass astrolabe blanks: should be 18 after Tuesday casting.\n- Mira said 2 were set aside for engraving practice, but I only saw one tagged practice blank.\n- Finished rete assemblies on cork board: 7 total? Wait, no — board has 6 hanging, plus 1 wrapped in linen in drawer N3.\n- Drawer N3 label says 'rete assembly, polished x2', but one bundle felt too light.\n\nLathe corner:\n- Ebony handles for demonstration pieces: 12 planned for the season.\n- Counted 9 in the tray at open.\n- Tomas borrowed 2 yesterday for photography props; not sure if returned.\n- Afternoon note says 'all handles back except cracked one' — unclear whether cracked one is included in the tray count or discarded.\n\nApprentice shelf:\n- Star charts, printed reference sheets: master set should be 24 sheets.\n- Current stack clipped in blue folder: 23 sheets.\n- One coffee-stained copy was removed last week, but Hana may have reprinted it.\n- There is also a loose duplicate of sheet 7 with handwritten angle corrections.\n\nFinishing room:\n- Lacquer jars, clear coat: purchase ledger says 8 jars delivered.\n- Visible on rack: 5 sealed, 1 open, 1 nearly empty.\n- Shipping crate by door still contains packing straw and maybe one jar, but nobody checked because the lid is jammed.\n- Safety log says minimum reserve before inspection is 6 usable jars.\n\nFront display prep:\n- Demonstration astrolabes ready for visitors: target 5.\n- Window stand currently holds 3.\n- One more complete unit on polishing cloth near the vice.\n- Another unit is missing its pointer pin; body and dial are assembled.\n\nTool roll inventory:\n- Fine gravers expected: 14.\n- Roll pockets occupied: 13.\n- One graver may be at sharpening station.\n- Sharpening station note says 'returned after honing', but timestamp smudged and no initials.\n\nPacking area:\n- Wool wrap squares for transport kits: expected 30.\n- Open bundle counted at 26.\n- Two kits already pre-packed with 2 wraps each.\n- Scrap bin contains 1 wrap cut into strips, not usable.\n\nGeneral remarks:\n- If an item is assembled but missing a single essential fastening piece, do not count it as inspection-ready.\n- Practice blanks should not count toward finished demonstration inventory.\n- Consumables that are open but still usable may count as usable reserve, but nearly empty containers should be flagged separately.\n- If two notes conflict, prefer the most recent note only when it is specific.\n\nWORKSHOP NOTE END\n\nReturn the answer in plain text with clear headings."
                }
            ]
        })


asyncio.run(main())

curl https://api.runware.ai/v1 \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "taskType": "textInference",
      "taskUUID": "ccb2bc2f-b5b7-4837-ae96-95e20990ba6f",
      "model": "anthropic:claude@haiku-4.5",
      "settings": {
        "systemPrompt": "You are a meticulous operations analyst. Resolve inconsistencies, preserve important details, and present findings in crisp sections with practical recommendations.",
        "maxTokens": 8192,
        "thinkingLevel": "medium",
        "cache": {
          "scope": "system",
          "ttl": "1h"
        }
      },
      "messages": [
        {
          "role": "user",
          "content": "You are helping a small historical-instrument workshop prepare for a guild inspection. Below is a messy internal note compiled from voice memos, sticky notes, and partial shift logs. Read it carefully, then do four things: (1) produce a clean inventory table with item name, expected count, confirmed count, status, and notes; (2) list every contradiction or ambiguity you found; (3) identify the top 5 issues that must be resolved before inspection, ranked by urgency; (4) draft a short message to the workshop lead summarizing what to fix today. Be conservative: if something is uncertain, say so explicitly.\n\nWORKSHOP NOTE START\n\nNorth bench cabinet:\n- Brass astrolabe blanks: should be 18 after Tuesday casting.\n- Mira said 2 were set aside for engraving practice, but I only saw one tagged practice blank.\n- Finished rete assemblies on cork board: 7 total? Wait, no — board has 6 hanging, plus 1 wrapped in linen in drawer N3.\n- Drawer N3 label says 'rete assembly, polished x2', but one bundle felt too light.\n\nLathe corner:\n- Ebony handles for demonstration pieces: 12 planned for the season.\n- Counted 9 in the tray at open.\n- Tomas borrowed 2 yesterday for photography props; not sure if returned.\n- Afternoon note says 'all handles back except cracked one' — unclear whether cracked one is included in the tray count or discarded.\n\nApprentice shelf:\n- Star charts, printed reference sheets: master set should be 24 sheets.\n- Current stack clipped in blue folder: 23 sheets.\n- One coffee-stained copy was removed last week, but Hana may have reprinted it.\n- There is also a loose duplicate of sheet 7 with handwritten angle corrections.\n\nFinishing room:\n- Lacquer jars, clear coat: purchase ledger says 8 jars delivered.\n- Visible on rack: 5 sealed, 1 open, 1 nearly empty.\n- Shipping crate by door still contains packing straw and maybe one jar, but nobody checked because the lid is jammed.\n- Safety log says minimum reserve before inspection is 6 usable jars.\n\nFront display prep:\n- Demonstration astrolabes ready for visitors: target 5.\n- Window stand currently holds 3.\n- One more complete unit on polishing cloth near the vice.\n- Another unit is missing its pointer pin; body and dial are assembled.\n\nTool roll inventory:\n- Fine gravers expected: 14.\n- Roll pockets occupied: 13.\n- One graver may be at sharpening station.\n- Sharpening station note says 'returned after honing', but timestamp smudged and no initials.\n\nPacking area:\n- Wool wrap squares for transport kits: expected 30.\n- Open bundle counted at 26.\n- Two kits already pre-packed with 2 wraps each.\n- Scrap bin contains 1 wrap cut into strips, not usable.\n\nGeneral remarks:\n- If an item is assembled but missing a single essential fastening piece, do not count it as inspection-ready.\n- Practice blanks should not count toward finished demonstration inventory.\n- Consumables that are open but still usable may count as usable reserve, but nearly empty containers should be flagged separately.\n- If two notes conflict, prefer the most recent note only when it is specific.\n\nWORKSHOP NOTE END\n\nReturn the answer in plain text with clear headings."
        }
      ]
    }
  ]'

runware run anthropic:claude@haiku-4.5 \
  settings.systemPrompt="You are a meticulous operations analyst. Resolve inconsistencies, preserve important details, and present findings in crisp sections with practical recommendations." \
  settings.maxTokens=8192 \
  settings.thinkingLevel=medium \
  settings.cache.scope=system \
  settings.cache.ttl=1h \
  messages.0.role=user \
  messages.0.content="You are helping a small historical-instrument workshop prepare for a guild inspection. Below is a messy internal note compiled from voice memos, sticky notes, and partial shift logs. Read it carefully, then do four things: (1) produce a clean inventory table with item name, expected count, confirmed count, status, and notes; (2) list every contradiction or ambiguity you found; (3) identify the top 5 issues that must be resolved before inspection, ranked by urgency; (4) draft a short message to the workshop lead summarizing what to fix today. Be conservative: if something is uncertain, say so explicitly.

WORKSHOP NOTE START

North bench cabinet:
- Brass astrolabe blanks: should be 18 after Tuesday casting.
- Mira said 2 were set aside for engraving practice, but I only saw one tagged practice blank.
- Finished rete assemblies on cork board: 7 total? Wait, no — board has 6 hanging, plus 1 wrapped in linen in drawer N3.
- Drawer N3 label says 'rete assembly, polished x2', but one bundle felt too light.

Lathe corner:
- Ebony handles for demonstration pieces: 12 planned for the season.
- Counted 9 in the tray at open.
- Tomas borrowed 2 yesterday for photography props; not sure if returned.
- Afternoon note says 'all handles back except cracked one' — unclear whether cracked one is included in the tray count or discarded.

Apprentice shelf:
- Star charts, printed reference sheets: master set should be 24 sheets.
- Current stack clipped in blue folder: 23 sheets.
- One coffee-stained copy was removed last week, but Hana may have reprinted it.
- There is also a loose duplicate of sheet 7 with handwritten angle corrections.

Finishing room:
- Lacquer jars, clear coat: purchase ledger says 8 jars delivered.
- Visible on rack: 5 sealed, 1 open, 1 nearly empty.
- Shipping crate by door still contains packing straw and maybe one jar, but nobody checked because the lid is jammed.
- Safety log says minimum reserve before inspection is 6 usable jars.

Front display prep:
- Demonstration astrolabes ready for visitors: target 5.
- Window stand currently holds 3.
- One more complete unit on polishing cloth near the vice.
- Another unit is missing its pointer pin; body and dial are assembled.

Tool roll inventory:
- Fine gravers expected: 14.
- Roll pockets occupied: 13.
- One graver may be at sharpening station.
- Sharpening station note says 'returned after honing', but timestamp smudged and no initials.

Packing area:
- Wool wrap squares for transport kits: expected 30.
- Open bundle counted at 26.
- Two kits already pre-packed with 2 wraps each.
- Scrap bin contains 1 wrap cut into strips, not usable.

General remarks:
- If an item is assembled but missing a single essential fastening piece, do not count it as inspection-ready.
- Practice blanks should not count toward finished demonstration inventory.
- Consumables that are open but still usable may count as usable reserve, but nearly empty containers should be flagged separately.
- If two notes conflict, prefer the most recent note only when it is specific.

WORKSHOP NOTE END

Return the answer in plain text with clear headings."

{
  "taskType": "textInference",
  "taskUUID": "ccb2bc2f-b5b7-4837-ae96-95e20990ba6f",
  "model": "anthropic:claude@haiku-4.5",
  "settings": {
    "systemPrompt": "You are a meticulous operations analyst. Resolve inconsistencies, preserve important details, and present findings in crisp sections with practical recommendations.",
    "maxTokens": 8192,
    "thinkingLevel": "medium",
    "cache": {
      "scope": "system",
      "ttl": "1h"
    }
  },
  "messages": [
    {
      "role": "user",
      "content": "You are helping a small historical-instrument workshop prepare for a guild inspection. Below is a messy internal note compiled from voice memos, sticky notes, and partial shift logs. Read it carefully, then do four things: (1) produce a clean inventory table with item name, expected count, confirmed count, status, and notes; (2) list every contradiction or ambiguity you found; (3) identify the top 5 issues that must be resolved before inspection, ranked by urgency; (4) draft a short message to the workshop lead summarizing what to fix today. Be conservative: if something is uncertain, say so explicitly.\n\nWORKSHOP NOTE START\n\nNorth bench cabinet:\n- Brass astrolabe blanks: should be 18 after Tuesday casting.\n- Mira said 2 were set aside for engraving practice, but I only saw one tagged practice blank.\n- Finished rete assemblies on cork board: 7 total? Wait, no — board has 6 hanging, plus 1 wrapped in linen in drawer N3.\n- Drawer N3 label says 'rete assembly, polished x2', but one bundle felt too light.\n\nLathe corner:\n- Ebony handles for demonstration pieces: 12 planned for the season.\n- Counted 9 in the tray at open.\n- Tomas borrowed 2 yesterday for photography props; not sure if returned.\n- Afternoon note says 'all handles back except cracked one' — unclear whether cracked one is included in the tray count or discarded.\n\nApprentice shelf:\n- Star charts, printed reference sheets: master set should be 24 sheets.\n- Current stack clipped in blue folder: 23 sheets.\n- One coffee-stained copy was removed last week, but Hana may have reprinted it.\n- There is also a loose duplicate of sheet 7 with handwritten angle corrections.\n\nFinishing room:\n- Lacquer jars, clear coat: purchase ledger says 8 jars delivered.\n- Visible on rack: 5 sealed, 1 open, 1 nearly empty.\n- Shipping crate by door still contains packing straw and maybe one jar, but nobody checked because the lid is jammed.\n- Safety log says minimum reserve before inspection is 6 usable jars.\n\nFront display prep:\n- Demonstration astrolabes ready for visitors: target 5.\n- Window stand currently holds 3.\n- One more complete unit on polishing cloth near the vice.\n- Another unit is missing its pointer pin; body and dial are assembled.\n\nTool roll inventory:\n- Fine gravers expected: 14.\n- Roll pockets occupied: 13.\n- One graver may be at sharpening station.\n- Sharpening station note says 'returned after honing', but timestamp smudged and no initials.\n\nPacking area:\n- Wool wrap squares for transport kits: expected 30.\n- Open bundle counted at 26.\n- Two kits already pre-packed with 2 wraps each.\n- Scrap bin contains 1 wrap cut into strips, not usable.\n\nGeneral remarks:\n- If an item is assembled but missing a single essential fastening piece, do not count it as inspection-ready.\n- Practice blanks should not count toward finished demonstration inventory.\n- Consumables that are open but still usable may count as usable reserve, but nearly empty containers should be flagged separately.\n- If two notes conflict, prefer the most recent note only when it is specific.\n\nWORKSHOP NOTE END\n\nReturn the answer in plain text with clear headings."
    }
  ]
}

import { createClient } from '@runware/sdk'

const client = await createClient({ apiKey: process.env.RUNWARE_API_KEY })
await client.connect()

const [result] = await client.run({
  model: 'anthropic:claude@haiku-4.5',
  settings: {
    maxTokens: 1200,
    thinkingLevel: 'medium'
  },
  messages: [
    {
      role: 'user',
      content: 'Act as a procurement analyst for an independent restaurant replacing a failed walk-in cooler. Prepare a concise decision memo for the owner using only the facts below.\n\nOperating constraints:\n- Total project budget: $48,000\n- Cooler must be operational within 6 calendar days of contract award\n- Off-site refrigerated storage costs $650 per outage day\n- The city requires a mechanical permit\n\nBid A — Northstar Mechanical\n- Base price: $41,800\n- Can start 9 days after award; installation takes 4 days\n- Includes demolition, disposal, insulated 8-by-12-foot box, condensing unit, installation, permit and up to 20 feet of electrical reconnection\n- Excludes night work\n- Warranty: 2 years labor, 5 years compressor, 10 years panels\n\nBid B — Polar Service Group\n- Base price: $37,900\n- Can start 3 days after award; installation takes 5 days\n- Includes insulated box, condensing unit and installation\n- Excludes permit ($1,200), electrical work (estimated at $4,800), and demolition/disposal ($2,600)\n- Change orders carry an 18% markup\n- Warranty: 1 year labor, 5 years compressor, 10 years panels\n\nBid C — Glacier Commercial Systems\n- Fixed price: $46,700\n- Can start 2 days after award; installation takes 3 days\n- Includes demolition, disposal, box, condensing unit, permit, electrical work, a temporary refrigerated trailer during the outage and a remote temperature alarm\n- Four-hour emergency service response\n- Warranty: 3 years labor, 7 years compressor, 10 years panels\n\nDeliver exactly these sections:\n1. Recommendation — no more than 80 words\n2. Normalized Comparison — a compact table showing quoted price, known add-ons, outage-storage cost, evaluated total, completion timing and warranty\n3. Key Risks — up to three bullets\n4. Pre-Award Questions — up to four questions for the recommended vendor\n5. Negotiation Priorities — up to three bullets\n\nShow arithmetic clearly. Treat the temporary trailer as eliminating off-site storage cost for Bid C. Flag any bid that fails a stated constraint. Do not invent taxes, financing costs or equipment specifications.'
    }
  ]
})

import asyncio
import os

from runware import Runware


async def main():
    async with Runware(api_key=os.environ["RUNWARE_API_KEY"]) as client:
        results = await client.run({
            "model": "anthropic:claude@haiku-4.5",
            "settings": {
                "maxTokens": 1200,
                "thinkingLevel": "medium"
            },
            "messages": [
                {
                    "role": "user",
                    "content": "Act as a procurement analyst for an independent restaurant replacing a failed walk-in cooler. Prepare a concise decision memo for the owner using only the facts below.\n\nOperating constraints:\n- Total project budget: $48,000\n- Cooler must be operational within 6 calendar days of contract award\n- Off-site refrigerated storage costs $650 per outage day\n- The city requires a mechanical permit\n\nBid A — Northstar Mechanical\n- Base price: $41,800\n- Can start 9 days after award; installation takes 4 days\n- Includes demolition, disposal, insulated 8-by-12-foot box, condensing unit, installation, permit and up to 20 feet of electrical reconnection\n- Excludes night work\n- Warranty: 2 years labor, 5 years compressor, 10 years panels\n\nBid B — Polar Service Group\n- Base price: $37,900\n- Can start 3 days after award; installation takes 5 days\n- Includes insulated box, condensing unit and installation\n- Excludes permit ($1,200), electrical work (estimated at $4,800), and demolition/disposal ($2,600)\n- Change orders carry an 18% markup\n- Warranty: 1 year labor, 5 years compressor, 10 years panels\n\nBid C — Glacier Commercial Systems\n- Fixed price: $46,700\n- Can start 2 days after award; installation takes 3 days\n- Includes demolition, disposal, box, condensing unit, permit, electrical work, a temporary refrigerated trailer during the outage and a remote temperature alarm\n- Four-hour emergency service response\n- Warranty: 3 years labor, 7 years compressor, 10 years panels\n\nDeliver exactly these sections:\n1. Recommendation — no more than 80 words\n2. Normalized Comparison — a compact table showing quoted price, known add-ons, outage-storage cost, evaluated total, completion timing and warranty\n3. Key Risks — up to three bullets\n4. Pre-Award Questions — up to four questions for the recommended vendor\n5. Negotiation Priorities — up to three bullets\n\nShow arithmetic clearly. Treat the temporary trailer as eliminating off-site storage cost for Bid C. Flag any bid that fails a stated constraint. Do not invent taxes, financing costs or equipment specifications."
                }
            ]
        })


asyncio.run(main())

curl https://api.runware.ai/v1 \
  -H "Authorization: Bearer $RUNWARE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '[
    {
      "taskType": "textInference",
      "taskUUID": "823d7dbd-68c5-408a-acab-02d7f7672f49",
      "model": "anthropic:claude@haiku-4.5",
      "settings": {
        "maxTokens": 1200,
        "thinkingLevel": "medium"
      },
      "messages": [
        {
          "role": "user",
          "content": "Act as a procurement analyst for an independent restaurant replacing a failed walk-in cooler. Prepare a concise decision memo for the owner using only the facts below.\n\nOperating constraints:\n- Total project budget: $48,000\n- Cooler must be operational within 6 calendar days of contract award\n- Off-site refrigerated storage costs $650 per outage day\n- The city requires a mechanical permit\n\nBid A — Northstar Mechanical\n- Base price: $41,800\n- Can start 9 days after award; installation takes 4 days\n- Includes demolition, disposal, insulated 8-by-12-foot box, condensing unit, installation, permit and up to 20 feet of electrical reconnection\n- Excludes night work\n- Warranty: 2 years labor, 5 years compressor, 10 years panels\n\nBid B — Polar Service Group\n- Base price: $37,900\n- Can start 3 days after award; installation takes 5 days\n- Includes insulated box, condensing unit and installation\n- Excludes permit ($1,200), electrical work (estimated at $4,800), and demolition/disposal ($2,600)\n- Change orders carry an 18% markup\n- Warranty: 1 year labor, 5 years compressor, 10 years panels\n\nBid C — Glacier Commercial Systems\n- Fixed price: $46,700\n- Can start 2 days after award; installation takes 3 days\n- Includes demolition, disposal, box, condensing unit, permit, electrical work, a temporary refrigerated trailer during the outage and a remote temperature alarm\n- Four-hour emergency service response\n- Warranty: 3 years labor, 7 years compressor, 10 years panels\n\nDeliver exactly these sections:\n1. Recommendation — no more than 80 words\n2. Normalized Comparison — a compact table showing quoted price, known add-ons, outage-storage cost, evaluated total, completion timing and warranty\n3. Key Risks — up to three bullets\n4. Pre-Award Questions — up to four questions for the recommended vendor\n5. Negotiation Priorities — up to three bullets\n\nShow arithmetic clearly. Treat the temporary trailer as eliminating off-site storage cost for Bid C. Flag any bid that fails a stated constraint. Do not invent taxes, financing costs or equipment specifications."
        }
      ]
    }
  ]'

runware run anthropic:claude@haiku-4.5 \
  settings.maxTokens=1200 \
  settings.thinkingLevel=medium \
  messages.0.role=user \
  messages.0.content="Act as a procurement analyst for an independent restaurant replacing a failed walk-in cooler. Prepare a concise decision memo for the owner using only the facts below.

Operating constraints:
- Total project budget: \$48,000
- Cooler must be operational within 6 calendar days of contract award
- Off-site refrigerated storage costs \$650 per outage day
- The city requires a mechanical permit

Bid A — Northstar Mechanical
- Base price: \$41,800
- Can start 9 days after award; installation takes 4 days
- Includes demolition, disposal, insulated 8-by-12-foot box, condensing unit, installation, permit and up to 20 feet of electrical reconnection
- Excludes night work
- Warranty: 2 years labor, 5 years compressor, 10 years panels

Bid B — Polar Service Group
- Base price: \$37,900
- Can start 3 days after award; installation takes 5 days
- Includes insulated box, condensing unit and installation
- Excludes permit (\$1,200), electrical work (estimated at \$4,800), and demolition/disposal (\$2,600)
- Change orders carry an 18% markup
- Warranty: 1 year labor, 5 years compressor, 10 years panels

Bid C — Glacier Commercial Systems
- Fixed price: \$46,700
- Can start 2 days after award; installation takes 3 days
- Includes demolition, disposal, box, condensing unit, permit, electrical work, a temporary refrigerated trailer during the outage and a remote temperature alarm
- Four-hour emergency service response
- Warranty: 3 years labor, 7 years compressor, 10 years panels

Deliver exactly these sections:
1. Recommendation — no more than 80 words
2. Normalized Comparison — a compact table showing quoted price, known add-ons, outage-storage cost, evaluated total, completion timing and warranty
3. Key Risks — up to three bullets
4. Pre-Award Questions — up to four questions for the recommended vendor
5. Negotiation Priorities — up to three bullets

Show arithmetic clearly. Treat the temporary trailer as eliminating off-site storage cost for Bid C. Flag any bid that fails a stated constraint. Do not invent taxes, financing costs or equipment specifications."

{
  "taskType": "textInference",
  "taskUUID": "823d7dbd-68c5-408a-acab-02d7f7672f49",
  "model": "anthropic:claude@haiku-4.5",
  "settings": {
    "maxTokens": 1200,
    "thinkingLevel": "medium"
  },
  "messages": [
    {
      "role": "user",
      "content": "Act as a procurement analyst for an independent restaurant replacing a failed walk-in cooler. Prepare a concise decision memo for the owner using only the facts below.\n\nOperating constraints:\n- Total project budget: $48,000\n- Cooler must be operational within 6 calendar days of contract award\n- Off-site refrigerated storage costs $650 per outage day\n- The city requires a mechanical permit\n\nBid A — Northstar Mechanical\n- Base price: $41,800\n- Can start 9 days after award; installation takes 4 days\n- Includes demolition, disposal, insulated 8-by-12-foot box, condensing unit, installation, permit and up to 20 feet of electrical reconnection\n- Excludes night work\n- Warranty: 2 years labor, 5 years compressor, 10 years panels\n\nBid B — Polar Service Group\n- Base price: $37,900\n- Can start 3 days after award; installation takes 5 days\n- Includes insulated box, condensing unit and installation\n- Excludes permit ($1,200), electrical work (estimated at $4,800), and demolition/disposal ($2,600)\n- Change orders carry an 18% markup\n- Warranty: 1 year labor, 5 years compressor, 10 years panels\n\nBid C — Glacier Commercial Systems\n- Fixed price: $46,700\n- Can start 2 days after award; installation takes 3 days\n- Includes demolition, disposal, box, condensing unit, permit, electrical work, a temporary refrigerated trailer during the outage and a remote temperature alarm\n- Four-hour emergency service response\n- Warranty: 3 years labor, 7 years compressor, 10 years panels\n\nDeliver exactly these sections:\n1. Recommendation — no more than 80 words\n2. Normalized Comparison — a compact table showing quoted price, known add-ons, outage-storage cost, evaluated total, completion timing and warranty\n3. Key Risks — up to three bullets\n4. Pre-Award Questions — up to four questions for the recommended vendor\n5. Negotiation Priorities — up to three bullets\n\nShow arithmetic clearly. Treat the temporary trailer as eliminating off-site storage cost for Bid C. Flag any bid that fails a stated constraint. Do not invent taxes, financing costs or equipment specifications."
    }
  ]
}