DeepSeek V3.2
deepseek/deepseek-v3.2
Compare with DeepSeek V3, DeepSeek V3.2 Exp, DeepSeek V3.1
CompareDeepSeek V3.2
| State | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Healthy | 100% | 97.60% | 94.24% | 94.24% | 858ms | 9.10s | 17 tps | 164K | $0.260 / $0.380 | |
DigitalOceandigitalocean | Healthy | 100% | 96.88% | 97.17% | 97.17% | 860ms | 7.78s | 44 tps | 164K | $0.250 / $0.800 |
Friendlifriendli | Healthy | 100% | 99.89% | 91.03% | 91.03% | 1.53s | 10.2s | 29 tps | 164K | $0.500 / $1.50 |
| Healthy | 100% | 99.50% | 99.54% | 99.54% | 2.19s | 8.36s | 23 tps | 164K | $0.209 / $0.310 | |
Googlegoogle-vertex | Healthy | 100% | 99.47% | 99.67% | 99.67% | 1.11s | 10.3s | 23 tps | 164K | $0.560 / $1.68 |
| Healthy | 100% | 99.98% | 99.99% | 99.99% | 1.29s | 5.44s | 25 tps | 164K | $0.269 / $0.400 | |
Venicevenice | Healthy | 100% | 82.18% | 75.39% | 75.39% | 1.91s | 10.2s | 13 tps | 160K | $0.330 / $0.480 |
| Healthy | 99.98% | 99.95% | 99.96% | 99.96% | 1.51s | 9.08s | 20 tps | 164K | $0.260 / $0.380 | |
| Healthy | 99.66% | 99.90% | 99.94% | 99.94% | 1.85s | 14.5s | 26 tps | 131K | $0.280 / $0.420 | |
| Healthy | 98.26% | 98.01% | 98.43% | 98.43% | 1.10s | 13.6s | 49 tps | 131K | $0.370 / $1.11 | |
| Degraded | 94.49% | 99.53% | 99.81% | 99.81% | 1.54s | 9.22s | 25 tps | 128K | $0.214 / $0.322 | |
| Degraded | 91.91% | 99.23% | 99.70% | 99.70% | 2.10s | 11.2s | 23 tps | 164K | $0.259 / $0.420 | |
Phalaphala | No data | — | 98.62% | 99.15% | 99.15% | 2.88s | 357s | 6.0 tps | 164K | $1.00 / $1.00 |
SambaNovasambanova | No data | — | 94.67% | 96.60% | 96.60% | 2.21s | 9.35s | 27 tps | 33K | $3.00 / $4.50 |
- DeepInfra100%Healthy24h 97.60%
- 7d
- 94.24%
- 30d
- 94.24%
- p50
- 858ms
- p99
- 9.10s
- Throughput
- 17 tps
- Context
- 164K
- $/Mtok
- $0.260 / $0.380
deepinfra/fp4fp4 - DigitalOcean100%Healthy24h 96.88%
- 7d
- 97.17%
- 30d
- 97.17%
- p50
- 860ms
- p99
- 7.78s
- Throughput
- 44 tps
- Context
- 164K
- $/Mtok
- $0.250 / $0.800
digitalocean - Friendli100%Healthy24h 99.89%
- 7d
- 91.03%
- 30d
- 91.03%
- p50
- 1.53s
- p99
- 10.2s
- Throughput
- 29 tps
- Context
- 164K
- $/Mtok
- $0.500 / $1.50
friendli - GMICloud100%Healthy24h 99.50%
- 7d
- 99.54%
- 30d
- 99.54%
- p50
- 2.19s
- p99
- 8.36s
- Throughput
- 23 tps
- Context
- 164K
- $/Mtok
- $0.209 / $0.310
gmicloud/fp8fp8 - Google100%Healthy24h 99.47%
- 7d
- 99.67%
- 30d
- 99.67%
- p50
- 1.11s
- p99
- 10.3s
- Throughput
- 23 tps
- Context
- 164K
- $/Mtok
- $0.560 / $1.68
google-vertex - Novita100%Healthy24h 99.98%
- 7d
- 99.99%
- 30d
- 99.99%
- p50
- 1.29s
- p99
- 5.44s
- Throughput
- 25 tps
- Context
- 164K
- $/Mtok
- $0.269 / $0.400
novita/fp8fp8 - Venice100%Healthy24h 82.18%
- 7d
- 75.39%
- 30d
- 75.39%
- p50
- 1.91s
- p99
- 10.2s
- Throughput
- 13 tps
- Context
- 160K
- $/Mtok
- $0.330 / $0.480
venice - AtlasCloud99.98%Healthy24h 99.95%
- 7d
- 99.96%
- 30d
- 99.96%
- p50
- 1.51s
- p99
- 9.08s
- Throughput
- 20 tps
- Context
- 164K
- $/Mtok
- $0.260 / $0.380
atlas-cloud/fp8fp8 - Baidu99.66%Healthy24h 99.90%
- 7d
- 99.94%
- 30d
- 99.94%
- p50
- 1.85s
- p99
- 14.5s
- Throughput
- 26 tps
- Context
- 131K
- $/Mtok
- $0.280 / $0.420
baidu/fp8fp8 - Alibaba98.26%Healthy24h 98.01%
- 7d
- 98.43%
- 30d
- 98.43%
- p50
- 1.10s
- p99
- 13.6s
- Throughput
- 49 tps
- Context
- 131K
- $/Mtok
- $0.370 / $1.11
alibaba/fp8fp8 - StreamLake94.49%Degraded24h 99.53%
- 7d
- 99.81%
- 30d
- 99.81%
- p50
- 1.54s
- p99
- 9.22s
- Throughput
- 25 tps
- Context
- 128K
- $/Mtok
- $0.214 / $0.322
streamlake/fp8fp8 - SiliconFlow91.91%Degraded24h 99.23%
- 7d
- 99.70%
- 30d
- 99.70%
- p50
- 2.10s
- p99
- 11.2s
- Throughput
- 23 tps
- Context
- 164K
- $/Mtok
- $0.259 / $0.420
siliconflow/fp8fp8 - No data24h 98.62%
- 7d
- 99.15%
- 30d
- 99.15%
- p50
- 2.88s
- p99
- 357s
- Throughput
- 6.0 tps
- Context
- 164K
- $/Mtok
- $1.00 / $1.00
phala - No data24h 94.67%
- 7d
- 96.60%
- 30d
- 96.60%
- p50
- 2.21s
- p99
- 9.35s
- Throughput
- 27 tps
- Context
- 33K
- $/Mtok
- $3.00 / $4.50
sambanova
Measured 4m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.
History
2 endpoints reporting no data are omitted from the charts.
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.
Throughput
Median tokens per second.