Artificial Analysis benchmarks open-weight gpt-oss-120b
Artificial Analysis has published a benchmark profile of gpt-oss-120b, an Apache 2.0 open-weight model, measuring its high-reasoning variant at about $0.35 and 7.4 minutes per Intelligence Index task.
Artificial Analysis has published a release page for gpt-oss-120b. The page reports intelligence, speed, latency and pricing for two configurations of the reasoning model, gpt-oss-120b (High) and gpt-oss-120b (Low).
The High setting is the stronger of the two. Its Intelligence Index score is listed as 4, and the page does not explain the scale. The High setting also has the lowest time per Intelligence Index task, at 7.4 minutes, and the lowest cost per task, at $0.35. The Low setting answers faster, with a time to first answer token of 2.28 seconds.
According to the listing, the model has:
- 117B total parameters, with 5.1B active per token
- a 131k-token context window
- text-only input and output
- a knowledge cutoff of May 31, 2024
- an Apache 2.0 license, with weights on Hugging Face
Artificial Analysis compares open-weight models only with other open-weight models of a similar size. At 117B parameters, gpt-oss-120b falls in its "Medium" class, which covers 40B to 150B parameters. Reasoning models like this one are also compared with non-reasoning models.
The page lists the evaluations behind its capability indexes. They include AA-Briefcase v1.1, AA-Omniscience v1.1, HLE-Diamond, AutomationBench-AA, GDP.pdf, MLCR-AA, CritPt v2 (Hard Subset), Terminal-Bench 4.0, Terminal-Bench-Science 0.1 and FrontierCode-AA 2.0. The extracted page doesn't include scores for the individual capability indexes or the model's rank against its peers.