Gemma 3 27B: providers compared
Every endpoint serving this model, side by side. ◆ marks the best value in each column — per model, never fleet-wide.
Current state
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100%◆ (best) | 99.10% | 99.18%◆ (best) | 99.47%◆ (best) | 1.94s | 8.22s | 18 tps | 131K | $0.080 / $0.160◆ (best) | |
| Healthy | 100%◆ (best) | 86.94% | 95.08% | 89.14% | 1.27s | 15.0s | 39 tps◆ (best) | 110K | $0.100 / $0.300 | |
| Healthy | 99.84% | 97.73% | 97.37% | 94.37% | 1.46s | 5.38s◆ (best) | 27 tps | 98K | $0.119 / $0.200 | |
| Healthy | 99.34% | 96.66% | 99.11% | 99.21% | 653ms◆ (best) | 6.33s | 39 tps◆ (best) | 131K | $0.080 / $0.450◆ (best) | |
Phalaphala | No data | — | — | — | — | — | — | — | 262K | $0.150 / $0.460 |
- DeepInfra100%◆ (best)Healthy24h 99.10%
- 7d
- 99.18%◆ (best)
- 30d
- 99.47%◆ (best)
- p50
- 1.94s
- p99
- 8.22s
- Throughput
- 18 tps
- Context
- 131K
- $/Mtok
- $0.080 / $0.160◆ (best)
deepinfra/fp8fp8 - Nebius100%◆ (best)Healthy24h 86.94%
- 7d
- 95.08%
- 30d
- 89.14%
- p50
- 1.27s
- p99
- 15.0s
- Throughput
- 39 tps◆ (best)
- Context
- 110K
- $/Mtok
- $0.100 / $0.300
nebius/fp8fp8 - Novita99.84%Healthy24h 97.73%
- 7d
- 97.37%
- 30d
- 94.37%
- p50
- 1.46s
- p99
- 5.38s◆ (best)
- Throughput
- 27 tps
- Context
- 98K
- $/Mtok
- $0.119 / $0.200
novita/bf16bf16 - Parasail99.34%Healthy24h 96.66%
- 7d
- 99.11%
- 30d
- 99.21%
- p50
- 653ms◆ (best)
- p99
- 6.33s
- Throughput
- 39 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.080 / $0.450◆ (best)
parasail/fp8fp8 - No data24h —
- 7d
- —
- 30d
- —
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 262K
- $/Mtok
- $0.150 / $0.460
phala
Measured 3m ago. Latency and throughput are OpenRouter's rolling 30-minute windows. ◆ marks the best value in each column among these endpoints.
Price vs latency
One dot per endpoint at the latest round · further left = faster, lower = cheaper · hover a dot for its tag
History
Availability
Axis starts at 90% so small differences stay visible; it expands to 0 when an endpoint falls below.
Time to first token
Logarithmic axis — the fleet spans two orders of magnitude. P50 is OpenRouter's rolling 30-minute window, averaged over the hour.
Throughput
Median tokens per second.