Guides

Examples

Common usage patterns for PromptDiff.

Choosing the cheapest model

Compare models and programmatically select the cheapest one that meets a quality threshold.

python

import promptdiff

client = promptdiff.Client(api_key="pd_...")

result = client.compare(
    prompt="Classify this text as spam or not: {input}",
    models=["gpt-4o-mini", "claude-3-haiku", "gemini-1.5-flash"],
    input="Congratulations! You have won a prize."
)

# Sort by cost
by_cost = sorted(result.results, key=lambda r: r.cost_usd)
cheapest = by_cost[0]
print("Cheapest:", cheapest.model, "cost:", cheapest.cost_usd)

Latency benchmarking

bash

curl -X POST https://promptdiff.bizmarq.com/api/v1/compare \
  -H "Authorization: Bearer pd_your_key" \
  -d '{
    "prompt": "Respond with exactly one word: yes",
    "models": ["gpt-4o-mini", "claude-3-haiku", "gemini-1.5-flash"],
    "options": { "max_tokens": 5 }
  }'

JavaScript / TypeScript

compare.ts

const response = await fetch(
  "https://promptdiff.bizmarq.com/api/v1/compare",
  {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      Authorization: `Bearer ${process.env.PROMPTDIFF_API_KEY}`,
    },
    body: JSON.stringify({
      prompt: "Explain closures in JavaScript briefly.",
      models: ["gpt-4o-mini", "claude-3-haiku"],
      options: { temperature: 0.5, max_tokens: 200 },
    }),
  }
);

const data = await response.json();
console.log(`Fastest: ${data.meta.fastest_model}`);
console.log(`Total cost: $${data.meta.total_cost_usd}`);