DeepSeek V4 Pro 0423: 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) | 97.56% | 98.37% | 98.37% | 1.71s | 17.3s | 37 tps | 1.0M | $1.42 / $2.83 | |
| Healthy | 100%◆ (best) | 95.11% | 96.76% | 96.76% | 1.35s | 62.2s | 18 tps | 1.0M | $1.68 / $3.38 | |
| Healthy | 100%◆ (best) | 99.06% | 98.03% | 98.03% | 840ms | 61.6s | 44 tps | 1.0M | $0.642 / $1.28 | |
| Healthy | 100%◆ (best) | 99.86% | 99.88%◆ (best) | 99.88%◆ (best) | 424ms | 2.81s◆ (best) | 60 tps | 1.0M | $1.74 / $3.48 | |
| Healthy | 100%◆ (best) | 99.55% | 98.71% | 98.71% | 792ms | 14.2s | 29 tps | 1.0M | $1.30 / $2.60 | |
| Healthy | 100%◆ (best) | 96.59% | 97.09% | 97.09% | 2.90s | 80.8s | 20 tps | 1.0M | $0.792 / $2.38 | |
| Healthy | 100%◆ (best) | 95.65% | 96.14% | 96.14% | 3.14s | 145s | 19 tps | 1.0M | $1.13 / $2.26 | |
| Healthy | 100%◆ (best) | 98.45% | 99.16% | 99.16% | 1.48s | 4.78s | 39 tps | 1.0M | $1.60 / $3.20 | |
| Healthy | 100%◆ (best) | 92.91% | 92.72% | 92.72% | 399ms◆ (best) | 55.0s | 57 tps | 1.0M | $1.74 / $3.48 | |
Togethertogether | Healthy | 100%◆ (best) | 96.21% | 97.59% | 97.59% | 594ms | 21.5s | 50 tps | 512K | $1.74 / $3.48 |
Venicevenice | Healthy | 100%◆ (best) | 82.76% | 84.13% | 84.13% | 1.11s | 75.8s | 27 tps | 1.0M | $1.65 / $3.30 |
| Healthy | 99.85% | 97.94% | 97.44% | 97.44% | 2.26s | 42.1s | 33 tps | 1.0M | $1.50 / $3.14 | |
Azureazure/us | Healthy | 99.82% | 94.83% | 94.61% | 94.61% | 3.08s | 45.2s | 39 tps | 1.0M | $1.91 / $3.83 |
DigitalOceandigitalocean | Healthy | 99.47% | 95.78% | 63.07% | 63.07% | 2.20s | 163s | 5.0 tps | 1.0M | $0.870 / $1.74 |
| Healthy | 99.16% | 98.93% | 98.36% | 98.36% | 2.22s | 17.5s | 26 tps | 1.0M | $0.634 / $1.27◆ (best) | |
| No data | — | 96.59% | 99.52% | 99.52% | 444ms | 48.8s | 57 tps | 1.0M | $1.15 / $2.55 | |
DeepSeekdeepseek | No data | — | — | 86.30% | 86.30% | — | — | — | 1.0M | $0.660 / $1.98 |
Fireworksfireworks | No data | — | 94.95% | 96.49% | 96.49% | 1.38s | 89.3s | 62 tps◆ (best) | 1.0M | $1.74 / $3.48 |
- Alibaba100%◆ (best)Healthy24h 97.56%
- 7d
- 98.37%
- 30d
- 98.37%
- p50
- 1.71s
- p99
- 17.3s
- Throughput
- 37 tps
- Context
- 1.0M
- $/Mtok
- $1.42 / $2.83
alibaba/fp8fp8 - AtlasCloud100%◆ (best)Healthy24h 95.11%
- 7d
- 96.76%
- 30d
- 96.76%
- p50
- 1.35s
- p99
- 62.2s
- Throughput
- 18 tps
- Context
- 1.0M
- $/Mtok
- $1.68 / $3.38
atlas-cloud/fp4fp4 - Baidu100%◆ (best)Healthy24h 99.06%
- 7d
- 98.03%
- 30d
- 98.03%
- p50
- 840ms
- p99
- 61.6s
- Throughput
- 44 tps
- Context
- 1.0M
- $/Mtok
- $0.642 / $1.28
baidu/fp8fp8 - BaseTen100%◆ (best)Healthy24h 99.86%
- 7d
- 99.88%◆ (best)
- 30d
- 99.88%◆ (best)
- p50
- 424ms
- p99
- 2.81s◆ (best)
- Throughput
- 60 tps
- Context
- 1.0M
- $/Mtok
- $1.74 / $3.48
baseten/fp4fp4 - DeepInfra100%◆ (best)Healthy24h 99.55%
- 7d
- 98.71%
- 30d
- 98.71%
- p50
- 792ms
- p99
- 14.2s
- Throughput
- 29 tps
- Context
- 1.0M
- $/Mtok
- $1.30 / $2.60
deepinfra/fp8fp8 - GMICloud100%◆ (best)Healthy24h 96.59%
- 7d
- 97.09%
- 30d
- 97.09%
- p50
- 2.90s
- p99
- 80.8s
- Throughput
- 20 tps
- Context
- 1.0M
- $/Mtok
- $0.792 / $2.38
gmicloud/fp8fp8 - Ionstream100%◆ (best)Healthy24h 95.65%
- 7d
- 96.14%
- 30d
- 96.14%
- p50
- 3.14s
- p99
- 145s
- Throughput
- 19 tps
- Context
- 1.0M
- $/Mtok
- $1.13 / $2.26
ionstream/fp4fp4 - Novita100%◆ (best)Healthy24h 98.45%
- 7d
- 99.16%
- 30d
- 99.16%
- p50
- 1.48s
- p99
- 4.78s
- Throughput
- 39 tps
- Context
- 1.0M
- $/Mtok
- $1.60 / $3.20
novita/fp8fp8 - Parasail100%◆ (best)Healthy24h 92.91%
- 7d
- 92.72%
- 30d
- 92.72%
- p50
- 399ms◆ (best)
- p99
- 55.0s
- Throughput
- 57 tps
- Context
- 1.0M
- $/Mtok
- $1.74 / $3.48
parasail/fp8fp8 - Together100%◆ (best)Healthy24h 96.21%
- 7d
- 97.59%
- 30d
- 97.59%
- p50
- 594ms
- p99
- 21.5s
- Throughput
- 50 tps
- Context
- 512K
- $/Mtok
- $1.74 / $3.48
together - Venice100%◆ (best)Healthy24h 82.76%
- 7d
- 84.13%
- 30d
- 84.13%
- p50
- 1.11s
- p99
- 75.8s
- Throughput
- 27 tps
- Context
- 1.0M
- $/Mtok
- $1.65 / $3.30
venice - SiliconFlow99.85%Healthy24h 97.94%
- 7d
- 97.44%
- 30d
- 97.44%
- p50
- 2.26s
- p99
- 42.1s
- Throughput
- 33 tps
- Context
- 1.0M
- $/Mtok
- $1.50 / $3.14
siliconflow/fp8fp8 - Azure99.82%Healthy24h 94.83%
- 7d
- 94.61%
- 30d
- 94.61%
- p50
- 3.08s
- p99
- 45.2s
- Throughput
- 39 tps
- Context
- 1.0M
- $/Mtok
- $1.91 / $3.83
azure/us - DigitalOcean99.47%Healthy24h 95.78%
- 7d
- 63.07%
- 30d
- 63.07%
- p50
- 2.20s
- p99
- 163s
- Throughput
- 5.0 tps
- Context
- 1.0M
- $/Mtok
- $0.870 / $1.74
digitalocean - StreamLake99.16%Healthy24h 98.93%
- 7d
- 98.36%
- 30d
- 98.36%
- p50
- 2.22s
- p99
- 17.5s
- Throughput
- 26 tps
- Context
- 1.0M
- $/Mtok
- $0.634 / $1.27◆ (best)
streamlake/fp8fp8 - No data24h 96.59%
- 7d
- 99.52%
- 30d
- 99.52%
- p50
- 444ms
- p99
- 48.8s
- Throughput
- 57 tps
- Context
- 1.0M
- $/Mtok
- $1.15 / $2.55
coreweave/fp8fp8 - No data24h —
- 7d
- 86.30%
- 30d
- 86.30%
- p50
- —
- p99
- —
- Throughput
- —
- Context
- 1.0M
- $/Mtok
- $0.660 / $1.98
deepseek - No data24h 94.95%
- 7d
- 96.49%
- 30d
- 96.49%
- p50
- 1.38s
- p99
- 89.3s
- Throughput
- 62 tps◆ (best)
- Context
- 1.0M
- $/Mtok
- $1.74 / $3.48
fireworks
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.
Throughput
Median tokens per second.