Sponsored by Byond Boundrys Consulting - Empowering Ideas, Delivering Results
OpenAI Advanced

gpt-oss-120b

gpt-oss-120b is OpenAI’s open-weight 117B (5.1B active) reasoning model that fits on a single H100 GPU, with a 131K context for private, self-hosted AI.

Reasoning ModelsText Free
In plain English

What is this model and why does it matter?

gpt-oss-120b is OpenAI's 117B-parameter, 5.1B-active open-weight reasoning model designed to fit on a single H100 GPU.

Self-hosted reasoningPrivate AIFine-tuningAgentic researchCommercial open-weight deployments
Model overview

gpt-oss-120b: features, use cases and important details

gpt-oss-120b is a current AI model verified from first-party OpenAI sources.

gpt-oss-120b verified specifications

gpt-oss-120b is OpenAI’s 117B-parameter, 5.1B-active open-weight reasoning model designed to fit on a single H100 GPU. Its verified context or usage limit is 131072 tokens, with 131072 tokens maximum output.

gpt-oss-120b pricing and access

OpenAI distributes gpt-oss-120b as Apache-2.0 open weights rather than a metered OpenAI API model; serving cost depends on your GPU infrastructure.

gpt-oss-120b best uses

Self-hosted reasoning, Private AI, Fine-tuning, Agentic research, Commercial open-weight deployments.

gpt-oss-120b limitations

Knowledge cutoff is June 2024, Deployment requires GPU operations, Local performance depends on inference stack.

Get started

How to use this model

  1. Download the official weights.
  2. Review Apache 2.0 terms.
  3. Provision suitable GPU hardware.
  4. Run a compatible inference stack.
  5. Tune reasoning effort and evaluate before production.
Copy and try

Example prompts

  • Analyze this codebase locally without sending data to an external API.
  • Fine-tune this model for a private domain.
  • Use tools to solve this structured reasoning task.
Capabilities

What it can do

  • 117B total / 5.1B active
  • 131K context
  • Apache 2.0
  • Configurable reasoning
  • Fine-tuning
  • Function calling
  • Structured outputs
Best for

Practical use cases

  • Private assistants
  • Research
  • Fine-tuning
  • Self-hosted agents
  • Regulated workloads
Pricing

What does it cost?

OpenAI distributes gpt-oss-120b as Apache-2.0 open weights rather than a metered OpenAI API model; serving cost depends on your GPU infrastructure.

InputSelf-hosted infrastructure cost
OutputSelf-hosted infrastructure cost
Simple summaryThere is no OpenAI per-token API price; users pay for their own GPU or hosting provider.

What stands out

  • Apache 2.0
  • Open weights
  • Fits on one H100
  • Fine-tunable

Things to consider

  • Infrastructure required
  • No hosted OpenAI API endpoint
  • Text-only
Limitations

Important restrictions and trade-offs

  • Knowledge cutoff is June 2024
  • Deployment requires GPU operations
  • Local performance depends on inference stack
SimplifyAITools verdict

Our editorial take

One of the most important open-weight OpenAI models for teams comparing private deployment with hosted frontier APIs.

References

Primary sources

  1. Open source 1 ↗
  2. Open source 2 ↗
Keep exploring

Related AI models