GPT OSS 120B
OpenAI's open-weight 120B model — the same architecture family as the GPT-5 line, runnable across many hosts.
Context window
128K
Max output
8K
Input price
$0.15 /1M
Output price
$0.60 /1M
Benchmarks
Independent benchmark scores — composite indices for reasoning, coding, and math, plus individual eval scores where available.
Intelligence Index33.3
Coding Index28.6
Math Index93.4
MMLU-Pro
81%
GPQA Diamond
78%
Humanity's Last Exam
19%
LiveCodeBench
88%
SciCode
39%
AIME 2025
93%
IFBench
69%
Long-context reasoning
51%
Terminal-Bench Hard
24%
τ²-Bench Telecom
66%
See full leaderboard →Benchmarks via Artificial Analysis · View on AA
Available routes
The same GPT OSS 120B runs on 8 different routes through the Opper gateway. Each route has its own data-handling posture.
| Provider | Region | Zero data retention | Training | Logging | Contract | Input | Output |
|---|---|---|---|---|---|---|---|
| AWS Bedrock | EU | Default | No | None | PAYG | $0.15 | $0.60 |
| Berget | EU | Default | No | None — "data is never stored or retained" per public materials | PAYG | $0.35 | $1.05 |
| Cerebras | US | Default | No | None by default per Privacy Policy. No documented abuse-monitoring window. | PAYG | $0.35 | $0.75 |
| Evroc | EU | None | No | Not enumerated publicly | PAYG | $0.22 | $0.86 |
| Fireworks | US | Default | No | None for open models by default — does not log or store prompt or generation data without explicit opt-in. | PAYG | $0.15 | $0.60 |
| Groq | US | Opt-in | No | None for inference by default. Up to 30 days for abuse-monitoring / system-reliability when triggered. | PAYG | $0.15 | $0.60 |
| Infercom | EU | — | No | Pending verification | PAYG | $0.24 | $0.64 |
| Nebius | EU | Opt-in | No | Per privacy policy: only as long as necessary; specific window not published. | PAYG | $0.15 | $0.60 |
Human review: No human review