midudev/isbetter.ai

⚔️ isbetter.ai

Compare AI models. Side by side.

Give every model the same prompt and compare the answer, the generated code, a live preview, speed, tokens and cost — all streaming in real time, right in your browser.

Live at isbetter.ai

Features

• Side-by-side battles — run one prompt against up to 6 models at once and watch every answer stream in.
• Three views per result — the raw output, the extracted code, and a sandboxed live preview of what the model built.
• Blind mode — hide and shuffle model identities so you judge the output, not the logo. Reveal whenever you're ready.
• Battle metrics — total time, time to first token, generation speed, tokens and estimated cost, with a timeline chart and per-metric winners.
• Shareable battles — publish a battle and send a link so anyone can review the results.
• Local history — every battle is saved in your browser so you can revisit or re-run it later.
• Bring your own keys — API keys are stored in localStorage and sent only to their provider. They never touch our servers.

Supported providers

| Provider | Notes | | --- | --- | | OpenRouter | Hundreds of models from every major lab through one key | | OpenAI | GPT models | | Anthropic | Claude models | | Google Gemini | Gemini models | | xAI | Grok models | | DeepSeek | DeepSeek chat and reasoner models | | Mistral AI | Mistral models | | Groq | Ultra-fast open-model inference | | Cerebras | Ultra-fast open-model inference | | Local | Any OpenAI-compatible server (Ollama, LM Studio, llama.cpp, …) |

Models served through an inference provider (OpenRouter, Groq, Cerebras, local) also show the badge of the lab that created the model.

Tech stack

• Astro with server routes on Cloudflare Workers
• Tailwind CSS 4 for styling
• Cloudflare D1 for share