Google: Gemma 3 27B
google/gemma-3-27b-it
Cheapest provider
$0.08 / 1M
DeepInfra
Fastest provider (p95)
58 tok/s
Parasail
Intelligence (composite)
66.3
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.
| Provider | Quant | Input $/1M | Output $/1M | Latency p50 / p95 | Throughput p50 / p95 | Uptime 30m | Success |
|---|---|---|---|---|---|---|---|
| DeepInfra | q8· fp8 | $0.0800 | $0.1600 | 916ms / 3916ms | 21 / 31 tok/s | 95.80% | 96.1% |
| Parasail | q8· fp8 | $0.0800 | $0.4500 | 1206ms / 3339ms | 31 / 58 tok/s | 100.00% | 99.5% |
| Nebius | q8· fp8 | $0.1000 | $0.3000 | 742ms / 2112ms | 19 / 48 tok/s | 99.87% | 99.9% |
| Phala | undisclosed | $0.1100 | $0.4000 | 810ms / 1728ms | 27 / 36 tok/s | 100.00% | 63.5% |
| Novita | full· bf16 | $0.1190 | $0.2000 | 2591ms / 29432ms | 15 / 29 tok/s | 98.34% | 97.9% |
“—” 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
64.3
broad reasoning
Code
79.2
pass@1 (HumanEval / LiveCodeBench)
MATH
75.2
math accuracy
GPQA Diamond
41.8
hard reasoning
Source: Gemma 3 technical report
How Atlas mode sources Google: Gemma 3 27B
- Strict mode — pin Google: Gemma 3 27B exactly and we pass it straight through, sourced from the cheapest provider above. The same model, no substitutions — currently DeepInfra at $0.08/1M.
- 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.