LLM Pricing API — AI Cost Arbitrage for Developers

How it works

Intelligent model routing in 3 steps

Our router finds the optimal model for your task — balancing cost, quality, and context window — so you don't have to.

1

Sign up free

Get your API key instantly. No credit card. No waitlist.

2

Set your constraints

Tell the router your max cost per 1M tokens, minimum quality score, or preferred provider.

3

Get routed automatically

The API returns the best model for your requirements with full pricing transparency.

4

Track your savings

See exactly how much you saved vs. the most expensive alternative considered.

# Find the best model under $2/1M tokens with quality ≥ 70
curl -X POST https://llm-pricing-api.fly.dev/router/recommend \
  -H "x-api-key: your-api-key" \
  -d '{"max_cost_per_1m_tokens": 2, "min_quality_score": 70}'

# Response
{
  "recommended": { "model_name": "gpt-4o-mini", "provider": "openai" },
  "score": 192.4,
  "reason": "Best quality/cost ratio; $0.375/1M tokens",
  "routing_id": "3f7a..."
}

Features

Everything you need to optimize AI spend

Built for developers and teams who care about cost without sacrificing quality.

🔀

Smart Router

Filter by cost, quality score, context window, provider, and task type. Get the optimal model every time.

📊

Live Pricing

Real-time prices from OpenAI, Anthropic, Google, DeepSeek, xAI, Cerebras, NVIDIA NIM, Groq, Mistral, Cohere, Together AI, and 10 more providers.

💰

Savings Tracker

Automatically tracks how much you save vs. the most expensive alternative on every routing decision.

⚡

Streaming Router

SSE streaming endpoint for real-time routing decisions. See every step of the recommendation process.

🏆

Quality Index

Models scored 0–100 on quality with a composite quality/value score so you never sacrifice capability.

🔌

OpenAI Compatible

Drop-in replacement for the OpenAI SDK. Point your base_url here — no code changes needed.

MCP Integration

Use from Claude, VS Code & Cursor

Connect any MCP-compatible AI assistant to your LLM Pricing server in under 2 minutes. No API key required — the server runs locally and imports pricing data directly.

Step 1 — Clone & install

git clone https://github.com/skakumanu/llm-pricing-mcp-server.git
cd llm-pricing-mcp-server
pip install -r requirements.txt

Step 2 — Add to your client config

📁 Windows: %APPDATA%\Claude\claude_desktop_config.json 📁 Mac: ~/Library/Application Support/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "llm-pricing": {
      "command": "python",
      "args": ["-u", "mcp/server.py"],
      "cwd": "/path/to/llm-pricing-mcp-server",
      "env": { "PYTHONPATH": "/path/to/llm-pricing-mcp-server", "PYTHONUNBUFFERED": "1" }
    }
  }
}

📁 Open Cline in VS Code → Settings (⚙️) → MCP Servers → Edit JSON

{
  "mcpServers": {
    "llm-pricing": {
      "command": "python",
      "args": ["-u", "mcp/server.py"],
      "cwd": "/path/to/llm-pricing-mcp-server",
      "env": { "PYTHONPATH": "/path/to/llm-pricing-mcp-server", "PYTHONUNBUFFERED": "1" }
    }
  }
}

📁 Windows: %USERPROFILE%\.cursor\mcp.json 📁 Mac/Linux: ~/.cursor/mcp.json

{
  "mcpServers": {
    "llm-pricing": {
      "command": "python",
      "args": ["-u", "mcp/server.py"],
      "cwd": "/path/to/llm-pricing-mcp-server",
      "env": { "PYTHONPATH": "/path/to/llm-pricing-mcp-server", "PYTHONUNBUFFERED": "1" }
    }
  }
}

Step 3 — Restart your client and ask

What's the cheapest model for RAG right now?

Compare GPT-4o vs Claude 3.5 Sonnet on cost and speed

What would 10M tokens cost on Gemini Flash?

Which model gives the best quality per dollar under $5/1M tokens?

29 tools available

get_all_pricing estimate_cost compare_costs get_performance_metrics get_pricing_trends get_pricing_history get_use_cases ask_agent register_price_alert list_price_alerts get_telemetry list_conversations delete_conversation delete_price_alert get_pricing_export_url

Pricing