OpenAI · Released Mar 10, 2026
gpt-oss-120b (high)
Open-source 120B model family offering clean open weights, high inference throughput, and easy fine-tuning hooks.
OpenRating Index
44.5
0–100 composite
Reasoning Index
61.8
0–100 composite
Cost / Task
$0.073
benchmark run
Output speed
164 tok/s
blended tok/s
Time to first token
0.4s
median latency
Input price
$0.06
per 1M tokens
Output price
$0.24
per 1M tokens
Context window
128K
tokens
Strengths
Open weights with Apache 2.0-style terms
Fast 164 tok/s throughput on commodity server nodes
Very low API serving overhead ($0.073/task)
How we rate gpt-oss-120b (high)
OpenRating scores every model on the same axes. The OpenRating Index is a weighted composite of reasoning, knowledge, and agentic benchmarks, normalised to a 0–100 scale. Speed is blended output measured in tokens per second under standardised load, with time-to-first-token reported separately. Prices are list input and output prices per million tokens, and Task Cost measures end-to-end workload spending.
OpenRating Index
44.5
Reasoning Index
61.8
Token Pricing
Input per 1M tokens
$0.06
Output per 1M tokens
$0.24
Cost of 1M input + 1M output
$0.3
Task Cost Economics
Empirical cost breakdown for a standard OpenRating Index benchmark task ($0.073 total).
Related models
Compare gpt-oss-120b (high) with the models it is most often measured against.