Batch Generation: Process Thousands of Prompts Simultaneously

When You Need Batch Processing

Data labeling, content generation at scale, bulk code analysis, test case generation — many real-world AI tasks involve processing hundreds or thousands of prompts. Sequential API calls are slow and error-prone. Batch Generation handles this efficiently.

How It Works

import vincony

client = vincony.Client(api_key="YOUR_KEY")

prompts = [
    {"role": "user", "content": f"Summarize: {doc}"}
    for doc in documents  # Your list of 1000+ documents
]

batch = client.batch.create(
    model="gemini-3-flash",
    messages_list=prompts,
    max_concurrent=50,
    on_progress=lambda p: print(f"{p.completed}/{p.total}")
)

# Wait for completion
results = batch.wait()
for r in results:
    print(r.choices[0].message.content[:100])

Features

• • Parallel processing: Up to 100 concurrent requests
• • Auto-retry: Failed requests are retried with exponential backoff
• • Progress tracking: Real-time completion stats via webhook or polling
• • Cost estimation: Get credit cost estimate before running
• • CSV export: Download results as CSV or JSON

Pricing

Batch processing uses the same per-model credit costs as individual requests — no batch surcharge. Available on Power ($54.99/mo) and Business ($199/mo) plans.