Qwen3.5 397B A17B: 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) | 99.72% | 99.09% | 97.58% | 1.09s | 6.24s | 32 tps | 262K | $0.450 / $3.00 | |
| Healthy | 100%◆ (best) | 99.85% | 98.47% | 99.11% | 681ms◆ (best) | 2.81s◆ (best) | 49 tps | 262K | $0.500 / $3.60 | |
Phalaphala | Healthy | 100%◆ (best) | 99.63% | 99.29% | 99.04% | 1.64s | 7.01s | 42 tps | 262K | $0.550 / $3.50 |
StreamLakestreamlake | Healthy | 100%◆ (best) | 99.05% | 98.87% | 97.50% | 1.71s | 8.00s | 43 tps | 256K | $0.600 / $3.60 |
Novitanovita | Healthy | 98.51% | 99.39% | 98.22% | 98.74% | 1.30s | 8.47s | 69 tps | 262K | $0.600 / $3.60 |
Alibabaalibaba | Healthy | 97.00% | 99.84% | 99.53%◆ (best) | 99.75%◆ (best) | 992ms | 4.28s | 95 tps◆ (best) | 262K | $0.390 / $2.34◆ (best) |
| No data | — | 95.63% | 97.64% | 97.46% | 1.22s | 9.18s | 61 tps | 262K | $0.550 / $3.50 | |
DigitalOceandigitalocean | No data | — | 94.63% | 93.61% | 93.76% | 1.25s | 9.02s | 8.0 tps | 131K | $0.550 / $3.50 |
| Degraded | — | 82.49% | 88.53% | 87.11% | 1.76s | 24.3s | 91 tps | 262K | $0.600 / $3.60 | |
| No data | — | — | — | — | — | — | — | 262K | $0.500 / $3.50 | |
Venicevenice | No data | — | 92.67% | 96.58% | 93.72% | 744ms | 14.8s | 64 tps | 128K | $0.750 / $4.50 |
- DeepInfra100%◆ (best)Healthy24h 99.72%
- 7d
- 99.09%
- 30d
- 97.58%
- p50
- 1.09s
- p99
- 6.24s
- Throughput
- 32 tps
- Context
- 262K
- $/Mtok
- $0.450 / $3.00
deepinfra/fp8fp8 - Parasail100%◆ (best)Healthy24h 99.85%
- 7d
- 98.47%
- 30d
- 99.11%
- p50
- 681ms◆ (best)
- p99
- 2.81s◆ (best)
- Throughput
- 49 tps
- Context
- 262K
- $/Mtok
- $0.500 / $3.60
parasail/fp8fp8 - Phala100%◆ (best)Healthy24h 99.63%
- 7d
- 99.29%
- 30d
- 99.04%
- p50
- 1.64s
- p99
- 7.01s
- Throughput
- 42 tps
- Context
- 262K
- $/Mtok
- $0.550 / $3.50
phala - StreamLake100%◆ (best)Healthy24h 99.05%
- 7d
- 98.87%
- 30d
- 97.50%
- p50
- 1.71s
- p99
- 8.00s
- Throughput
- 43 tps
- Context
- 256K
- $/Mtok
- $0.600 / $3.60
streamlake - Novita98.51%Healthy24h 99.39%
- 7d
- 98.22%
- 30d
- 98.74%
- p50
- 1.30s
- p99
- 8.47s
- Throughput
- 69 tps
- Context
- 262K
- $/Mtok
- $0.600 / $3.60
novita - Alibaba97.00%Healthy24h 99.84%
- 7d
- 99.53%◆ (best)
- 30d
- 99.75%◆ (best)
- p50
- 992ms
- p99
- 4.28s
- Throughput
- 95 tps◆ (best)
- Context
- 262K
- $/Mtok
- $0.390 / $2.34◆ (best)
alibaba - No data24h 95.63%
- 7d
- 97.64%
- 30d
- 97.46%
- p50
- 1.22s
- p99
- 9.18s
- Throughput
- 61 tps
- Context
- 262K
- $/Mtok
- $0.550 / $3.50
atlas-cloud/fp8fp8 - No data24h 94.63%
- 7d
- 93.61%
- 30d
- 93.76%
- p50
- 1.25s
- p99
- 9.02s
- Throughput
- 8.0 tps
- Context
- 131K
- $/Mtok
- $0.550 / $3.50
digitalocean - Degraded24h 82.49%
- 7d
- 88.53%
- 30d
- 87.11%
- p50
- 1.76s
- p99
- 24.3s
- Throughput
- 91 tps
- Context
- 262K
- $/Mtok
- $0.600 / $3.60
gmicloud/fp8fp8 - No data24h —
- 7d
- —
- 30d
- —
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 262K
- $/Mtok
- $0.500 / $3.50
morph/fp4fp4 - No data24h 92.67%
- 7d
- 96.58%
- 30d
- 93.72%
- p50
- 744ms
- p99
- 14.8s
- Throughput
- 64 tps
- Context
- 128K
- $/Mtok
- $0.750 / $4.50
venice
Measured 5m 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.