OpenAI: GPT-OSS 120B
openai/gpt-oss-120b
Try GPT-OSS 120B 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: "openai/gpt-oss-120b"
}).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: "openai/gpt-oss-120b"
}).then(response => {
document.body.innerHTML = response.message.content;
});
</script>
</body>
</html>
Model Card
GPT-OSS 120B is OpenAI's most powerful open-weight model under Apache 2.0 license, achieving near-parity with o4-mini on reasoning benchmarks. It has 117B total parameters with 5.1B active, fitting on a single H100 GPU.
Context Window 131K
tokens
Max Output 131K
tokens
Input Cost $0.04
per million tokens
Output Cost $0.17
per million tokens
Release Date Aug 5, 2025
Output Speed 212
tokens / sec
Latency 0.50s
time to first token
Try GPT-OSS 120B for free
Try GPT-OSS 120B instantly in your browser.
This playground uses the Puter.js AI API — no API keys or setup required.
Benchmarks
How GPT-OSS 120B performs on standard evaluations.
| Benchmark | Score |
|---|---|
| GPQA Diamond Graduate-level science Q&A | 78.2% |
| Humanity's Last Exam Cross-domain reasoning | 19.6% |
| LiveCodeBench Recent coding problems | 87.8% |
| SciCode Scientific programming | 34.0% |
| AIME 2025 Advanced math exam | 93.4% |
| IFBench Instruction following | 69.0% |
| LCR Long-context reasoning | 52.0% |
| Terminal-Bench Hard Agentic terminal tasks | 23.5% |
| τ²-Bench Tool use / agents | 65.8% |
Scores sourced from Artificial Analysis.
Find other OpenAI models →
GPT-6.1 Sol
GPT-6.1 Sol is OpenAI's mid-tier model in the GPT-6 family, released September 29, 2026 as an update to GPT-6 Sol. It sits below the flagship GPT-6 Astra, takes text and image input, supports tool calling, and has a 1,050,000-token context window with 128,000 max output tokens. Pricing is unchanged from GPT-6 Sol at $2 per million input tokens and $10 per million output tokens, while scores are higher. It reaches 75.2% on DeepSWE 1.1 (68.8% for GPT-6 Sol), 71.4% on OSWorld 2.0 (64.4%), and 32.0% on GDP.pdf (28.0%). Its factual error rate at low reasoning effort fell from 11.4% to 7.7%. In one third-party comparison table, GPT-6 Astra scores 74.8% on DeepSWE and 73.5% on OSWorld at roughly five times the price. GPT-6.1 Sol suits coding agents, computer use, document question answering, and workflow automation where cost per task matters.
ChatGPT-6.1 Sol Pro
GPT-6.1 Sol Pro is OpenAI's higher-effort configuration of GPT-6.1 Sol, the mid-tier model in the GPT-6 family. Based on how OpenAI handled GPT-6 Sol Pro, it is likely the same underlying model with reasoning mode set to pro, though OpenAI's documentation for the 6.1 release does not confirm this. It has a 1,050,000-token context window and 128,000 max output tokens, at $2 per million input tokens and $10 per million output tokens. Supported parameters include reasoning effort, tools, tool choice, structured response formats, seed, and streaming. Sol Pro is intended for complex requests such as agentic coding, computer use, and workflow automation, where answer quality matters more than latency. We found no published benchmarks for the Pro configuration itself.
ChatGPT-6 Sol
GPT-6 Sol is OpenAI's mid-tier model in the GPT-6 family, released September 22, 2026 as a faster, cheaper counterpart to the flagship GPT-6 Astra, built on the same training methods. OpenAI positions it for complex development work that developers run repeatedly, such as building features, reviewing code, debugging, and analyzing data. OpenAI reports Sol scoring 68.8% on the DeepSWE 1.1 coding benchmark, 60.5% on OSWorld 2.0 for computer use at xhigh effort, and 56.4% on Agents' Last Exam, a test of professional work across 55 sub-industries. The company says Sol makes about half as many factual errors as GPT-5.6 Sol, approaching Astra-level reliability at a lower cost. At $2 per million input tokens and $10 per million output tokens, half of GPT-5.6 Sol's rate, Sol matches Claude Sonnet 5's standard pricing, though no published benchmark compares the two models directly.
Frequently Asked Questions
You can access GPT-OSS 120B by OpenAI 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.
GPT-OSS 120B 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.
GPT-OSS 120B 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 | $0.04 |
| Output | $0.17 |
GPT-OSS 120B was created by OpenAI and released on Aug 5, 2025.
GPT-OSS 120B supports a context window of 131K tokens. For reference, that is roughly equivalent to 262 pages of text.
GPT-OSS 120B can generate up to 131K tokens in a single response.
GPT-OSS 120B scores 11.6 on the Artificial Analysis Intelligence Index, outperforming 47% of tracked models. On coding, it scores 30.4 (outperforms 36% of models). On math, it scores 93.4 (outperforms 94% of models).
Yes — the GPT-OSS 120B 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 GPT-OSS 120B to your app for free
Developers can integrate GPT-OSS 120B for free using the Puter.js AI API.
With the User-Pays Model, each user covers their own AI usage instead of the developer.