NEWMEN
← back to all models

OpenAI: gpt-oss-120b

openai/gpt-oss-120b

OpenAIclosed-weight131K context19 providersIntelligence 81.3

Cheapest provider

$0.04 / 1M

DeepInfra

Fastest provider (p95)

2152 tok/s

Cerebras

Intelligence (composite)

81.3

MMLU-Pro · HumanEval · math · GPQA

Per-provider performance

Latency / throughput / uptime / price measured across providers over the last 30 minutes of live traffic. This is what proves “sourced cheapest” — Atlas mode draws on these per call to serve the cheapest path that holds quality.

ProviderQuantInput $/1MOutput $/1MLatency p50 / p95Throughput p50 / p95Uptime 30mSuccess
DeepInfrafull· bf16$0.0390$0.1900657ms / 10623ms82 / 405.8 tok/s98.66%59.0%
DekaLLMfull· bf16$0.0390$0.1800509ms / 1530ms5 / 39.1 tok/s96.09%78.9%
Novitaq4· fp4$0.0500$0.2500450ms / 1027ms97 / 198 tok/s100.00%100.0%
SiliconFlowq8· fp8$0.0500$0.45001416ms / 4011ms17 / 38 tok/s92.56%92.2%
Googlefull· unknown$0.0900$0.3600314ms / 833ms221 / 658 tok/s100.00%100.0%
BaseTenq4· fp4$0.1000$0.5000185ms / 891ms252 / 397 tok/s100.00%100.0%
DigitalOceanfull· unknown$0.1000$0.7000513ms / 1506ms127 / 162 tok/s99.72%99.7%
Parasailq4· fp4$0.1000$0.7500319ms / 655ms252 / 430.2 tok/s100.00%99.9%
SambaNovafull· unknown$0.1400$0.9500486ms / 4269ms177 / 430 tok/s97.83%85.2%
Amazon Bedrockfull· unknown$0.1500$0.600098.53%
Amazon Bedrockfull· unknown$0.1500$0.6000
DeepInfrafull· bf16$0.1500$0.6000657ms / 10623ms82 / 405.8 tok/s98.18%59.0%
Groqfull· unknown$0.1500$0.6000104ms / 649ms443 / 855.4 tok/s99.99%100.0%
Marafull· unknown$0.1500$0.7500943ms / 2413ms89.5 / 460.9 tok/s99.10%78.6%
Nebiusq4· fp4$0.1500$0.6000207ms / 1221ms308 / 489.3 tok/s100.00%100.0%
Phalafull· unknown$0.1500$0.60001156ms / 2222ms91 / 122 tok/s100.00%100.0%
Togetherfull· unknown$0.1500$0.6000307ms / 1501ms62 / 112 tok/s99.80%99.9%
WandBq4· fp4$0.1500$0.6000274ms / 565ms50 / 114 tok/s100.00%100.0%
Cerebrasfull· fp16$0.3500$0.7500196ms / 652ms694 / 2152 tok/s99.97%99.4%

“—” means live telemetry hasn’t accumulated enough recent traffic for that endpoint. “undisclosed” means the provider serves the model but doesn’t expose the quantization label (typically running fp8 / int8 internally).

Intelligence breakdown

Composite score is a weighted average of public benchmarks (30% MMLU-Pro, 25% code pass@1, 25% math, 20% GPQA). Numbers come from model cards and the Artificial Analysis intelligence harness; missing components are renormalised over what’s present.

MMLU-Pro

broad reasoning

Code

71.0

pass@1 (HumanEval / LiveCodeBench)

AIME 2025

92.5

math accuracy

GPQA Diamond

80.1

hard reasoning

Source: OpenAI gpt-oss model card (GPQA-Diamond 80.1, AIME25 92.5 no-tools, HumanEval ~71); arxiv.org/abs/2508.10925

How Atlas mode sources OpenAI: gpt-oss-120b

  • Strict mode — pin OpenAI: gpt-oss-120b exactly and we pass it straight through, sourced from the cheapest provider above. The same model, no substitutions — currently DeepInfra at $0.04/1M.
  • Atlas mode — the default. Each call is auto-optimized for the cheapest path that holds quality, at least 5% off going direct from call one and climbing as it ramps. You always see which model served the call and exactly what you saved — thumbs-down anything you don’t like for a full refund.