Unbiased: Pareto
unbiased/pareto
Try Pareto for free in your browser, and add it to your app for free with Puter.js AI API.
Try it free Add to your app// npm install @heyputer/puter.js
import { puter } from '@heyputer/puter.js';
puter.ai.chat("Explain quantum computing in simple terms", {
model: "unbiased/pareto"
}).then(response => {
document.body.innerHTML = response.message.content;
});
<html>
<body>
<script src="https://js.puter.com/v2/"></script>
<script>
puter.ai.chat("Explain quantum computing in simple terms", {
model: "unbiased/pareto"
}).then(response => {
document.body.innerHTML = response.message.content;
});
</script>
</body>
</html>
Model Card
Pareto is a blended AI model from Unbiased AI, built by Circuit & Chisel, a remote-first team based in the US and Canada. Rather than routing a request to one model, it runs several models against the same request and returns the best answer, without switching mid-conversation, which preserves prompt caching.
On Unbiased's published benchmarks, Pareto 26.9 scores 88/100 on ArXivMath and 78/100 on MMMU-Pro, and ties GPT 6 Astra and DeepSeek 4.1 Flash at 74/100 on the DeepSWE coding benchmark. It trails Fable 5.1 and GPT 6 Astra on Terminal-Bench 4.0 and HLE, and Unbiased publishes those losses alongside its wins rather than only citing favorable results.
It accepts text and image input and supports tool calling. It targets engineering teams evaluating model performance, platform teams looking for a cost-effective general-purpose option, and developers building agents that need consistent responses across a session.
Context Window 1M
tokens
Max Output 131K
tokens
Input Cost $2.5
per million tokens
Output Cost $7.5
per million tokens
Tool Use Yes
Release Date Sep 17, 2026
Try Pareto for free
Try Pareto instantly in your browser.
This playground uses the Puter.js AI API — no API keys or setup required.
More AI Models From Unbiased
Pareto 26.10
Pareto 26.10 is a composite AI model from Unbiased AI, built by Circuit & Chisel, a remote-first team in the US and Canada. It runs several models on the same request and returns one answer, and it is aimed at research, coding, and agentic workflows. Parallel's search leaderboard lists it with a Search Intelligence Score of 68.6 (34.8 without search) at $116 per 1,000 tasks, from an evaluation dated October 1, 2026. Scores for the earlier Pareto 26.10 Preview were run by Unbiased and have not been independently reproduced. It accepts text and image input, supports tool calling, and has a 1,048,576-token context window with up to 131,072 output tokens. It suits engineering teams, platform teams, and agent builders who want a general-purpose model with a low cost per task.
ChatPareto 26.9
Pareto 26.9 is a blended AI model from Unbiased AI, the platform of Circuit & Chisel, a remote-first team in the US and Canada. It runs several frontier and open models on each request and returns one answer. It does not switch models within a conversation, so prompt caching still applies. On Unbiased's published scores, Pareto 26.9 gets 74 on DeepSWE, 88 on ArXivMath, and 78 on MMMU-Pro. It scores 51 on Terminal-Bench 4.0 and 49 on Humanity's Last Exam without tools, which are below the frontier models. Because the underlying model mix can change between releases, scores cannot be attributed to a single model. It supports tool calling and has a 262,144-token context window with up to 131,072 output tokens. It suits coding and agent workflows that need consistent behavior across a session.
ChatPareto 26.10 Preview
Pareto 26.10 Preview is a composite AI model from Unbiased AI, built by Circuit & Chisel, a remote-first team in the US and Canada. It runs several models on the same request and returns one answer, and it is aimed at research, coding, and agentic workflows. On Unbiased's preliminary, vendor-run benchmarks it scores 69.9% on DeepSWE v1.1 at $0.24 per task, 50.8% on Terminal-Bench 4.0, 49.9% on Humanity's Last Exam (text only), and 92.4% on GPQA-Diamond. On DeepSWE it is level with Fable 5 (70.0%, $13.50 per task). On Terminal-Bench, GPT 6 Astra (58) and Claude Fable 5.1 (56) score higher. It accepts text and image input, supports tool calling, and has a 1,048,576-token context window with up to 131,072 output tokens. It suits engineering teams, platform teams, and agent builders who want a low cost per task.
Frequently Asked Questions
You can access Pareto by Unbiased through Puter.js AI API. Include the library in your web app or Node.js project and start making calls with just a few lines of JavaScript — no backend and no configuration required. You can also use it with Python or cURL via Puter's OpenAI-compatible API.
Pareto is free to try with a Puter account. Every account includes a free AI allowance, and you can chat with it in the playground on this page. You can upgrade your account anytime for a larger allowance.
Pareto is free to integrate using the Puter.js AI API. With the User-Pays Model, you can add AI to your app for $0, since users cover their own AI usage through their Puter account.
| Price per 1M tokens | |
|---|---|
| Input | $2.5 |
| Output | $7.5 |
Pareto was created by Unbiased and released on Sep 17, 2026.
Pareto supports a context window of 1M tokens. For reference, that is roughly equivalent to 2,097 pages of text.
Pareto can generate up to 131K tokens in a single response.
Yes, Pareto supports tool use (function calling), allowing it to interact with external tools, APIs, and data sources as part of its response flow.
Yes — the Pareto API works with any JavaScript framework, Node.js, or plain HTML through Puter.js. Just include the library and start building. See the documentation for more details.
Add Pareto to your app for free
Developers can integrate Pareto for free using the Puter.js AI API.
With the User-Pays Model, each user covers their own AI usage instead of the developer.