NEW

Flint-Smart: Multi-Step Problem Solving Engine

Not just another API router. Flint-Smart decomposes complex problems, executes them in parallel, and synthesizes coherent answers — all behind a single endpoint.

Try in Playground How It Works

Pipeline: 5 Stages to a Better Answer

1

Analyze

Fast rule-based complexity check — simple, medium, or complex

2

Decompose

LLM breaks the problem into MECE subtasks with rule-based fallback

3

Execute

Parallel execution across specialized models — code, reasoning, chat

4

Synthesize

Premium model weaves results into one coherent answer

5

Verify

Confidence scoring with automatic iteration when needed

One Endpoint. Smarter Answers.

cURL — No model parameter needed 
curl https://api.flintapi.ai/v1/smart/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "messages": [
      {"role": "user", "content": "Compare React vs Vue for building an e-commerce platform. Consider performance, ecosystem, learning curve, and SEO."}
    ]
  }'

Flint-Smart automatically detects complexity, decomposes the comparison into parallel research subtasks, and synthesizes a structured comparison — all behind a single call.

What You Get Back

{
  "model": "flint-smart",
  "choices": [{
    "message": {
      "role": "assistant",
      "content": "## Overview\nBoth React and Vue are excellent choices..."
    }
  }],
  "_orchestration": {
    "strategy": "multi-step",
    "complexity": "complex",
    "steps": 4,
    "models_used": ["deepseek-v4-pro", "qwen3.5-flash"],
    "elapsed_ms": 3200
  }
}

Why Flint-Smart vs. Others

Feature Flint-Smart OpenRouter OpenAI
Auto model selection ✔ Built-in Manual Single model
Multi-step decomposition ✔ LLM + rule-based ✘ ✘
Parallel execution ✔ asyncio.gather ✘ ✘
Confidence verification ✔ Auto-iterate ✘ ✘
Chinese LLMs ✔ 22+ models Few ✘
OpenAI-compatible API ✔ Drop-in ✔ ✔
Price (1M tokens) From $0.14 From $0.30 From $0.15
Response metadata ✔ Per-model billing Basic Basic

When to Use Flint-Smart

🔍 Research & Comparison

"Compare X vs Y for Z use case" — automatically researches both sides in parallel and synthesizes a balanced analysis.

💻 Code + Explanation

"Write a Python API with JWT auth and explain the design" — code generation and architecture explanation run in parallel.

📊 Strategy & Planning

"Create a go-to-market plan for a SaaS product" — decomposes into market analysis, competitor research, and tactical planning.

🛠️ Architecture Design

"Design a microservices architecture for a food delivery app" — splits into data modeling, API design, and infrastructure planning.