{
"name": "Compare answers, speed and cost of Claude, GPT, Gemini, DeepSeek and Qwen side by side with apimodels.app",
"nodes": [
{
"parameters": {
"content": "## Compare chat models side by side: answers, speed and cost\n\nPick a few models, type one prompt, and get every answer on one page with how long each took, how many tokens it used and what it actually cost. Useful when choosing a model for a feature, checking a prompt across vendors, or showing a team why one model is worth the price. All models run through one OpenAI-compatible key on [apimodels.app](https://apimodels.app).\n\n### How it works\n1. The form takes a prompt, an optional system message, 2 to 6 models and an output limit.\n2. **One item per model** builds one request per selected model.\n3. **Ask each model** sends them all at once to the chat completions endpoint.\n4. **Get billing record** looks up each request by its id: the exact amount charged, token counts, and start and finish times measured on the server.\n5. The result page shows a comparison table and every full answer, and marks the fastest and the lowest-cost model.\n\n### Setup\n1. Create an API key in the [apimodels console](https://apimodels.app/console/api-keys). New accounts get $0.1 of free credit.\n2. Create a **Bearer Auth** credential with that key and select it in the two HTTP Request nodes.\n3. Activate the workflow and open the production form URL from the trigger node.\nStep-by-step guide: [apimodels.app/docs/n8n](https://apimodels.app/docs/n8n)\n\n### Customize\n- Add any model id from `GET /v1/models` to the list in **One item per model** and to the form options.\n- Reasoning models spend part of the output limit on thinking; raise the limit if answers stop early.",
"height": 860,
"width": 540,
"color": 1
},
"id": "fc0e1dc7-3515-58ed-a11f-995f4bcd457a",
"name": "Overview",
"type": "n8n-nodes-base.stickyNote",
"typeVersion": 1,
"position": [
-660,
-320
]
},
{
"parameters": {
"content": "## 1. Ask every model\nOne request per selected model, all sent at the same time.",
"height": 400,
"width": 900,
"color": 7
},
"id": "bf3ff6f0-11ce-5e10-a995-53dd26250c22",
"name": "Section: ask",
"type": "n8n-nodes-base.stickyNote",
"typeVersion": 1,
"position": [
-40,
20
]
},
{
"parameters": {
"content": "## 2. Measure and compare\nBilling records give the real cost and server-side timing; the page puts it all in one table.",
"height": 400,
"width": 680,
"color": 7
},
"id": "e45fd964-2bf0-5b2a-a012-8132afe1443d",
"name": "Section: measure",
"type": "n8n-nodes-base.stickyNote",
"typeVersion": 1,
"position": [
900,
20
]
},
{
"parameters": {
"formTitle": "Compare chat models",
"formDescription": "Send one prompt to several models and compare their answers, speed and cost.",
"formFields": {
"values": [
{
"fieldLabel": "Prompt",
"fieldType": "textarea",
"requiredField": true,
"placeholder": "e.g. Write a 3-sentence product description for a smart water bottle."
},
{
"fieldLabel": "System message",
"fieldType": "textarea",
"placeholder": "Optional, e.g. You are a concise copywriter."
},
{
"fieldLabel": "Models",
"fieldType": "checkbox",
"requiredField": true,
"limitSelection": "range",
"minSelections": 2,
"maxSelections": 6,
"fieldOptions": {
"values": [
{
"option": "Claude Sonnet 5.5"
},
{
"option": "Claude Haiku 5.5"
},
{
"option": "GPT-5.5"
},
{
"option": "GPT-6 Luna"
},
{
"option": "Gemini 4 Argon"
},
{
"option": "Gemini 3.8 Flash"
},
{
"option": "DeepSeek V4.1 Flash"
},
{
"option": "Qwen3.8 Max"
},
{
"option": "Grok 4.6"
},
{
"option": "GLM-5.3"
}
]
}
},
{
"fieldLabel": "Output limit (tokens)",
"fieldType": "dropdown",
"requiredField": true,
"fieldOptions": {
"values": [
{
"option": "2000"
},
{
"option": "4000"
},
{
"option": "8000"
}
]
}
}
]
},
"options": {
"path": "compare-models",
"buttonLabel": "Compare"
}
},
"id": "48ccaffe-b353-588d-a16c-b86bc5656faa",
"name": "Comparison form",
"type": "n8n-nodes-base.formTrigger",
"typeVersion": 2.2,
"position": [
0,
240
],
"webhookId": "4fbaec26-26db-5ac8-94cf-dab91afae8e6"
},
{
"parameters": {
"assignments": {
"assignments": [
{
"id": "66e95071-5a71-5423-889d-9f089661e572",
"name": "api_base",
"value": "https://api.apimodels.app/v1",
"type": "string"
}
]
},
"includeOtherFields": true,
"options": {}
},
"id": "36c1d1bb-a2e3-5eaa-9ac7-afbdb6bf2fa7",
"name": "Settings",
"type": "n8n-nodes-base.set",
"typeVersion": 3.4,
"position": [
220,
240
]
},
{
"parameters": {
"jsCode": "// Form label -> model id. Any id from GET /v1/models works.\nconst MODELS = {\n 'Claude Sonnet 5.5': 'claude-sonnet-5-5',\n 'Claude Haiku 5.5': 'claude-haiku-5-5',\n 'GPT-5.5': 'gpt-5.5',\n 'GPT-6 Luna': 'gpt-6-luna',\n 'Gemini 4 Argon': 'gemini-4-argon',\n 'Gemini 3.8 Flash': 'gemini-3.8-flash',\n 'DeepSeek V4.1 Flash': 'deepseek-v4.1-flash',\n 'Qwen3.8 Max': 'qwen3.8-max',\n 'Grok 4.6': 'grok-4.6',\n 'GLM-5.3': 'glm-5.3',\n};\n\nconst form = $input.first().json;\nconst picked = Array.isArray(form['Models']) ? form['Models'] : [form['Models']];\nconst messages = [];\nif ((form['System message'] || '').trim()) messages.push({ role: 'system', content: form['System message'].trim() });\nmessages.push({ role: 'user', content: form['Prompt'] });\n\nreturn picked.filter((label) => MODELS[label]).map((label) => ({\n json: {\n label,\n body: { model: MODELS[label], max_tokens: Number(form['Output limit (tokens)']) || 2000, messages },\n },\n}));"
},
"id": "135f4537-e02e-5c4f-9363-d2087c4891ff",
"name": "One item per model",
"type": "n8n-nodes-base.code",
"typeVersion": 2,
"position": [
440,
240
]
},
{
"parameters": {
"method": "POST",
"url": "={{ $('Settings').first().json.api_base }}/chat/completions",
"authentication": "genericCredentialType",
"genericAuthType": "httpBearerAuth",
"sendBody": true,
"specifyBody": "json",
"jsonBody": "={{ JSON.stringify($json.body) }}",
"options": {
"response": {
"response": {
"fullResponse": true
}
},
"timeout": 240000
}
},
"id": "f6dda213-7964-5137-b7b0-bee69c36941e",
"name": "Ask each model",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
660,
240
],
"credentials": {
"httpBearerAuth": {
"id": "REPLACE_WITH_YOUR_CREDENTIAL",
"name": "apimodels API key"
}
},
"onError": "continueRegularOutput"
},
{
"parameters": {
"url": "={{ $('Settings').first().json.api_base }}/records/{{ ($json.headers && $json.headers['x-apimodels-request-id']) || 'none' }}",
"authentication": "genericCredentialType",
"genericAuthType": "httpBearerAuth",
"options": {
"timeout": 30000
}
},
"id": "fab41ca4-59b0-50a3-ac93-84e467495cbb",
"name": "Get billing record",
"type": "n8n-nodes-base.httpRequest",
"typeVersion": 4.2,
"position": [
960,
240
],
"credentials": {
"httpBearerAuth": {
"id": "REPLACE_WITH_YOUR_CREDENTIAL",
"name": "apimodels API key"
}
},
"onError": "continueRegularOutput",
"retryOnFail": true,
"maxTries": 3,
"waitBetweenTries": 2000
},
{
"parameters": {
"jsCode": "const esc = (s) => String(s ?? '').replace(/&/g, '&').replace(//g, '>').replace(/\"/g, '"');\nconst reqs = $('One item per model').all();\nconst answers = $('Ask each model').all();\nconst records = $input.all();\nconst form = $('Comparison form').first().json;\n\nconst rows = reqs.map((r, i) => {\n const a = answers[i] ? answers[i].json : {};\n const rec = (records[i] && records[i].json && records[i].json.data) || {};\n const text = a.body && a.body.choices && a.body.choices[0] && a.body.choices[0].message ? a.body.choices[0].message.content : null;\n const err = text == null ? ((a.error && (a.error.message || a.error)) || 'No answer') : null;\n const seconds = rec.created_at && rec.completed_at ? (rec.completed_at - rec.created_at) / 1000 : null;\n const usage = rec.usage || {};\n return {\n label: r.json.label, model: r.json.body.model, text, err, seconds,\n cost: typeof rec.credits === 'number' ? rec.credits : null,\n input: usage.input_tokens, output: usage.output_tokens,\n stop: a.body && a.body.choices && a.body.choices[0] ? a.body.choices[0].finish_reason : null,\n };\n});\n\nconst ok = rows.filter((r) => !r.err);\nconst fastest = ok.filter((r) => r.seconds != null).sort((x, y) => x.seconds - y.seconds)[0];\nconst costs = ok.filter((r) => r.cost != null).map((r) => r.cost);\nconst lowestCost = costs.length ? Math.min(...costs) : null; // ties all get the badge\nconst total = rows.reduce((s, r) => s + (r.cost || 0), 0);\nconst badge = (r) => [r === fastest ? 'fastest' : '', (r.cost != null && r.cost === lowestCost) ? 'lowest cost' : ''].join(' ');\n\nconst table = rows.map((r) => `${esc(r.label)}
${esc(r.model)} ${badge(r)} | \n${r.err ? '—' : (r.seconds != null ? r.seconds.toFixed(1) + ' s' : '?')} | \n${r.err ? '—' : `${r.input ?? '?'} / ${r.output ?? '?'}`} | \n${r.err ? '$0 (failed)' : (r.cost != null ? '$' + r.cost.toFixed(4) : '?')} |
`).join('');\n\nconst cards = rows.map((r) => `${esc(r.label)} ${badge(r)}${r.stop === 'length' ? ' (stopped at the output limit)' : ''}
\n${r.err ? '' + esc(r.err) + '' : esc(r.text)}
Model comparison
\n
${esc(form['Prompt'])}
\n
| Model | Time | Tokens in / out | Charged |
|---|
${table}
\n
Time is measured on the server from request to last token. Charged is the exact amount from GET /v1/records/{id}; failed calls are not charged. Total: $${total.toFixed(4)}.
\n${cards}\n
Compare again