gpt-oss-120b

reasoningapache-2.0120B

`gpt-oss-120b`is our most powerful open-weight model, which fits into a single H100 GPU (117B parameters with 5.1B active parameters).

Intelligence
Reasoning
Speed
Context
131K
Max output
131K

Capabilities

vision
tool call
structured output
reasoning
json mode
streaming
fine tuning
batch

Details

Provider OpenAI
Creator openai
Familygpt-oss
Licenseapache-2.0
Parameters120B
Statusactive
Input modalitiestext
Output modalitiestext
Architecture—
Knowledge cutoff
Training data cutoff—
Release date—
Deprecation date—
Typereasoning
Reasoning tokensYes
Max input—
Open weightYes
Sourceofficial
Last updated

Tools

Function Callingfunction_calling
Call external functions and APIs
Code Interpretercode_interpreter
Execute code in a sandbox
MCPmcp
Connect to Model Context Protocol servers
Web Searchweb_search
Search the web for information

Endpoints

ResponsesPOST
Stateful multi-turn conversationshttps://api.openai.com/v1/responses
BatchPOST
Async batch processing at lower costhttps://api.openai.com/v1/batches

Price across providers

ProviderInputOutputSince launch
DeepInfracheapest$0.04$0.17↓5%
OpenRoutercheapest$0.04$0.17↓5%
Cloudflare AI Gateway$0.04$0.19—
Novita AI$0.05$0.25—
Groq$0.15$0.6—
SambaNova Cloud$0.22$0.59—
Cerebras$0.35$0.75—
Vercel AI Gateway$0.35$0.75—

Family Comparison: gpt-oss

ModelContextMax outPricing
gpt-oss-120b131K131K
gpt-oss-20b131K131K

API

GET/v1/models/openai/gpt-oss-120b