- [Build a real-time fact checker with Parallel and Cerebras](https://parallel.ai/blog/cerebras-fact-checker)
Tags:Cookbook
Reading time: 5 min# Introducing research models with Basis for the Parallel Chat API
Tags:Product Release
Reading time: 2 min
Starting today, the Parallel Chat API supports three new research models, bringing research-grade web intelligence with full Basis verification[Basis verification]($https://docs.parallel.ai/task-api/guides/access-research-basis) to interactive AI applications.
The Chat API now offers two modes: the speed model for low-latency responses across a broad range of use cases, and research models (Lite, Base, Core) for deeper outputs where comprehensive verification matters more than milliseconds. Both return OpenAI ChatCompletions-compatible streaming text and JSON.
12345678910111213141516171819202122232425from openai import OpenAI
client = OpenAI(
api_key="PARALLEL_API_KEY",
base_url="https://api.parallel.ai"
)
response = client.chat.completions.create(
model="core", # Research model with full Basis support
messages=[
{"role": "user", "content": "What are the key risk factors for Acme Corp?"}
],
)
print(response.choices[0].message.content)
print(response.basis) # Citations, reasoning, confidence scores``` from openai import OpenAI client = OpenAI( api_key="PARALLEL_API_KEY", base_url="https://api.parallel.ai" ) response = client.chat.completions.create( model="core", # Research model with full Basis support messages=[ {"role": "user", "content": "What are the key risk factors for Acme Corp?"} ], ) print(response.choices[0].message.content) print(response.basis) # Citations, reasoning, confidence scores``` ## More models for different needs
The speed model delivers ~3-second time-to-first-token for chat interfaces and interactive tools. Research models trade latency for depth, running the same processors that power our Task API, now accessible through an OpenAI-compatible interface.
## Research-grade verification with Basis
Every response from research models includes Basis, the same verification framework trusted in production workflows across insurance, finance, and sales intelligence.
Basis provides four components for every output: citations linking to source URLs, reasoning explaining how conclusions were reached, excerpts containing relevant text from sources, and calibrated confidence scores classified as low, medium, or high.
123456789101112131415161718192021{
"field": "risk_factors",
"citations": [
{
"url": "https://sec.gov/filings/acme-10k-2024",
"excerpts": ["The company faces significant supply chain concentration risk..."]
}
],
"reasoning": "Multiple SEC filings and analyst reports corroborate supply chain dependencies...",
"confidence": "high"
}``` { "field": "risk_factors", "citations": [ { "url": "https://sec.gov/filings/acme-10k-2024", "excerpts": ["The company faces significant supply chain concentration risk..."] } ], "reasoning": "Multiple SEC filings and analyst reports corroborate supply chain dependencies...", "confidence": "high" }``` These confidence scores are calibrated against real-world datasets. High-confidence outputs have 2-3x lower error rates than overall dataset averages. Low-confidence flags identify exactly where human review adds value, enabling efficient workflows that focus attention only where it matters.
## **Why research models matter**
Research models give the Chat API flexibility to think more deeply and search the web more widely. When a question requires multi-hop reasoning, synthesis across scattered sources, or verification from primary documents, research models allocate the compute and retrieval budget to get the best answer, not just the fastest one.
## **Start building**
Research models are available today in the Parallel Chat API. Get started in our Developer Platform[Developer Platform]($https://platform.parallel.ai/) or dive into the documentation[documentation]($https://docs.parallel.ai/).
## **About Parallel Web Systems**
Parallel develops critical web search infrastructure for AI. Our suite of web search and agent APIs is built on a rapidly growing proprietary index of the global internet. These solutions transform human tasks that previously took days and weeks into agentic tasks that now take seconds and minutes.
Fortune 100 and 500 companies use Parallel's web intelligence APIs in insurance, finance, and retail workflows to automate critical business functions. Leading AI-native businesses like Starbridge, Amp, and Day AI use Parallel to support core features like public sector contract monitoring, documentation lookup, and GTM operations.
By Parallel
January 15, 2026
