OpenAI: gpt-oss-120b
openai/gpt-oss-120b
Cheapest provider
$0.04 / 1M
DeepInfra
Fastest provider (p95)
2613 tok/s
Cerebras
Intelligence (composite)
81.3
MMLU-Pro · HumanEval · math · GPQA
Per-provider performance
Latency / throughput / uptime measured across providers over the last 30 minutes of live traffic. Atlas’s router weighs these per call (with the eval-gate signal) when picking a variant for Standard and Batch tiers.
| Provider | Quant | Input $/1M | Output $/1M | Latency p50 / p95 | Throughput p50 / p95 | Uptime 30m | Success |
|---|---|---|---|---|---|---|---|
| DeepInfra | full· bf16 | $0.0390 | $0.1900 | 409ms / 10066ms | 220 / 445.7 tok/s | 100.00% | 42.7% |
| DekaLLM | full· bf16 | $0.0390 | $0.1800 | 588ms / 3369ms | 9 / 36 tok/s | 98.95% | 97.9% |
| Novita | q4· fp4 | $0.0500 | $0.2500 | 554ms / 1613ms | 33 / 228 tok/s | 99.79% | 100.0% |
| SiliconFlow | q8· fp8 | $0.0500 | $0.4500 | 1981ms / 10515ms | 9 / 20 tok/s | 58.43% | 100.0% |
| full· unknown | $0.0900 | $0.3600 | 692ms / 2165ms | 97 / 154 tok/s | 99.93% | 100.0% | |
| BaseTen | q4· fp4 | $0.1000 | $0.5000 | 242ms / 844ms | 214 / 317 tok/s | 99.95% | 99.1% |
| Parasail | q4· fp4 | $0.1000 | $0.7500 | 460ms / 1170ms | 76 / 95 tok/s | 100.00% | 93.8% |
| Phala | full· unknown | $0.1000 | $0.4900 | 983ms / 1831ms | 79 / 110 tok/s | 99.94% | 99.6% |
| SambaNova | full· unknown | $0.1400 | $0.9500 | 677ms / 3183ms | 136 / 677 tok/s | 99.58% | 70.2% |
| Amazon Bedrock | full· unknown | $0.1500 | $0.6000 | 1546ms / 7909ms | 157 / 241 tok/s | 100.00% | 100.0% |
| Amazon Bedrock | full· unknown | $0.1500 | $0.6000 | 1546ms / 7909ms | 157 / 241 tok/s | 25.60% | 100.0% |
| Ambient | full· unknown | $0.1500 | $0.6000 | 483ms / 1478ms | 72 / 138 tok/s | 99.96% | 97.9% |
| DeepInfra | full· bf16 | $0.1500 | $0.6000 | 409ms / 10066ms | 220 / 445.7 tok/s | 99.75% | 42.7% |
| Groq | full· unknown | $0.1500 | $0.6000 | 170ms / 1000ms | 382.5 / 756 tok/s | 99.97% | 98.5% |
| Mara | full· unknown | $0.1500 | $0.7500 | 640ms / 3004ms | 140 / 475.6 tok/s | 99.94% | 83.6% |
| Nebius | q4· fp4 | $0.1500 | $0.6000 | 233ms / 10121ms | 107 / 221 tok/s | 100.00% | 95.5% |
| Together | full· unknown | $0.1500 | $0.6000 | 780ms / 2323ms | 65 / 126 tok/s | 99.10% | 99.8% |
| WandB | q4· fp4 | $0.1500 | $0.6000 | 408ms / 1014ms | 118 / 134 tok/s | 100.00% | 100.0% |
| Cerebras | full· fp16 | $0.3500 | $0.7500 | 194ms / 711ms | 990 / 2612.9 tok/s | 100.00% | 96.4% |
“—” 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 routes OpenAI: gpt-oss-120b
- Realtime tier — direct passthrough at the upstream’s native precision. Best for hard latency / quality guarantees.
- Standard tier — Atlas picks the cheapest provider variant whose quantization has stayed green on your operation’s eval gates. For OpenAI: gpt-oss-120b that’s currently DeepInfra at $0.04/1M.
- Batch tier — async, biggest discount. Roadmapped to use provider batch APIs (OpenAI / Anthropic) where available and queued spot capacity for open-weight workloads.