Qwen3.5-27B: 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) | 99.72% | 99.80%◆ (best) | 99.80%◆ (best) | 1.24s | 4.82s | 19 tps | 262K | $0.300 / $2.40 | |
| Healthy | 100%◆ (best) | 99.15% | 98.99% | 98.99% | 1.27s | 7.36s | 10 tps | 262K | $0.250 / $2.00 | |
| Healthy | 96.15% | 99.30% | 99.02% | 99.02% | 754ms | 6.24s | 17 tps | 262K | $0.270 / $2.16 | |
Alibabaalibaba | Down | 83.16% | 97.34% | 97.97% | 97.97% | 1.14s | 16.9s | 18 tps | 262K | $0.195 / $1.56◆ (best) |
| Down | 82.89% | 98.68% | 97.91% | 97.91% | 312ms◆ (best) | 1.87s◆ (best) | 41 tps◆ (best) | 262K | $0.260 / $2.60 | |
Phalaphala | No data | — | 77.20% | 65.66% | 65.66% | 1.75s | 10.1s | 18 tps | 262K | $0.300 / $2.40 |
- Novita100%◆ (best)Healthy24h 99.72%
- 7d
- 99.80%◆ (best)
- 30d
- 99.80%◆ (best)
- p50
- 1.24s
- p99
- 4.82s
- Throughput
- 19 tps
- Context
- 262K
- $/Mtok
- $0.300 / $2.40
novita/bf16bf16 - SiliconFlow100%◆ (best)Healthy24h 99.15%
- 7d
- 98.99%
- 30d
- 98.99%
- p50
- 1.27s
- p99
- 7.36s
- Throughput
- 10 tps
- Context
- 262K
- $/Mtok
- $0.250 / $2.00
siliconflow/fp8fp8 - AtlasCloud96.15%Healthy24h 99.30%
- 7d
- 99.02%
- 30d
- 99.02%
- p50
- 754ms
- p99
- 6.24s
- Throughput
- 17 tps
- Context
- 262K
- $/Mtok
- $0.270 / $2.16
atlas-cloud/fp8fp8 - Alibaba83.16%Down24h 97.34%
- 7d
- 97.97%
- 30d
- 97.97%
- p50
- 1.14s
- p99
- 16.9s
- Throughput
- 18 tps
- Context
- 262K
- $/Mtok
- $0.195 / $1.56◆ (best)
alibaba - DeepInfra82.89%Down24h 98.68%
- 7d
- 97.91%
- 30d
- 97.91%
- p50
- 312ms◆ (best)
- p99
- 1.87s◆ (best)
- Throughput
- 41 tps◆ (best)
- Context
- 262K
- $/Mtok
- $0.260 / $2.60
deepinfra/fp8fp8 - No data24h 77.20%
- 7d
- 65.66%
- 30d
- 65.66%
- p50
- 1.75s
- p99
- 10.1s
- Throughput
- 18 tps
- Context
- 262K
- $/Mtok
- $0.300 / $2.40
phala
Measured 2m 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.