GPT-OSS 120B

Summary: GPT-OSS 120B is a cutting-edge open-source Mixture of Experts (MoE) model, specifically optimized for agent workflows and complex reasoning tasks. This medium-sized model combines the efficiency of selective parameter activation with exceptional language understanding, making it ideal for research, development, and production environments where transparency, customization, and sophisticated AI capabilities are essential.

Intelligence	Speed	Sovereignty	Input	Output
High	Medium	Medium	Text	Text

Central parameters

Description: Open-source Mixture-of-Experts architecture with efficient expert routing for optimized inference performance.

Model identifier: openai/gpt-oss-120b

IONOS AI Model Hub Lifecycle and Alternatives

IONOS Launch	End of Life	Alternative	Successor
August 12, 2025	N/A	Llama 3.3 70B

Origin

Provider	Country	License	Flavor	Release
OpenAI	USA	Apache 2.0	Base	August 2025

Technology

Context window	Parameters	Quantization	Multilingual	Further details
128k	120B	MXFP4	Yes	Hugging Face

Modalities

Text	Image	Audio
Input and output	Not supported	Not supported

Endpoints

Chat Completions	Embeddings	Image generation
v1/chat/completions	Not supported	Not supported

Features

Streaming	Reasoning	Tool calling
Supported	Supported	Supported

Reasoning Example

GPT-OSS 120B supports advanced reasoning capabilities with configurable reasoning effort. Control how deeply the model thinks by setting the reasoning_effort parameter to low, medium (default), or high. The model's reasoning process is included in the output response.

Reasoning Effort Levels:

• Low: Fast responses with minimal internal reasoning. Best for straightforward questions and when speed is prioritized. Uses fewer tokens.

• Medium (default): Balanced approach with moderate reasoning depth. Suitable for most use cases requiring thoughtful responses.

• High: Deep analytical thinking with extensive internal reasoning. Ideal for complex problem-solving, mathematical proofs, and multi-step reasoning tasks. Uses more tokens due to an extended reasoning process.

Higher reasoning effort levels result in more comprehensive analysis but consume additional tokens and increase response time. The reasoning tokens are included in token usage.

Request

{
  "stream": false,
  "model": "openai/gpt-oss-120b",
  "reasoning_effort": "low",
  "messages": [
    {
      "role": "user",
      "content": "Answer me with one letter. Maybe A."
    }
  ]
}

Response (shortened for readability)

{
  "choices": [
    {
      "message": {
        "role": "assistant",
        "content": "A",
        "reasoning": "User asks to answer with one letter, maybe A. So respond with a single letter."
      }
    }
  ]
}

Rate limits

Rate limits ensure fair usage and reliable access to the AI Model Hub. In addition to the contract-wide rate limits, no model-specific limits apply.