Re-benchmarked weekly — MiMo, DeepSeek, Gemini & more
Stop paying GPT-4o prices for “hi”.
Model Router is a drop-in OpenAI-compatible proxy. It classifies every prompt and routes it to the cheapest model that still clears your quality bar — saving Indian dev teams 60-80% on LLM spend with one line of code.
Read the 2-min quickstart · no credit card · bring your own keys · change one base URL
ts
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://router.aiskillhub.info/api/v1", // ← one line
apiKey: process.env.MODEL_ROUTER_KEY,
});
// model: "auto" lets the router pick the cheapest capable model
const res = await client.chat.completions.create({
model: "auto",
messages: [{ role: "user", content: "Summarize this PR in one line" }],
});The live routing table
We continuously re-benchmark cheap models so your routing stays optimal — no code changes.
| Model | Input $/1M | Output $/1M | Quality | Re-benchmarked |
|---|---|---|---|---|
| MiMo 7B / xiaomi | $0.04 | $0.12 | 58 | 2026-05-24 |
| DeepSeek V3 / deepseek | $0.14 | $0.28 | 71 | 2026-05-20 |
| Gemini 2.5 Flash / google | $0.15 | $0.60 | 74 | 2026-05-18 |
| GPT-4o mini / openai | $0.15 | $0.60 | 72 | 2026-05-15 |
| DeepSeek R1 / deepseekreasoning | $0.55 | $2.19 | 84 | 2026-05-20 |
| Claude Haiku 4.5 / anthropic | $0.80 | $4.00 | 81 | 2026-05-15 |
| GPT-4o / openai | $2.50 | $10.00 | 86 | 2026-05-15 |
| Claude Sonnet 4.6 / anthropicreasoning | $3.00 | $15.00 | 92 | 2026-05-15 |
Built for the price war
“We were sending every support reply to GPT-4o. Switched the base URL, kept our quality, and our monthly bill dropped from $2,100 to $480.”
— Design partner, 12-person dev agency, Bengaluru
Simple, India-first pricing
Start free. Upgrade when you outgrow it.