DeepSeek V4 Pro 0813: 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 | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
Alibabaalibaba | Healthy | 100%◆ (best) | 99.82% | 99.62% | 99.62% | 1.29s | 7.62s | 54 tps | 1.0M | $0.603 / $1.81◆ (best) |
| Healthy | 100%◆ (best) | 98.93% | 98.67% | 98.67% | 1.54s | 16.9s | 103 tps◆ (best) | 1.0M | $1.32 / $3.96 | |
Cloudflarecloudflare | Healthy | 100%◆ (best) | 98.09% | 97.21% | 97.21% | 4.62s | 47.1s | 37 tps | 1.0M | $1.32 / $3.96 |
DeepSeekdeepseek | Healthy | 100%◆ (best) | 99.95% | 86.81% | 86.81% | 1.64s | 9.50s | 31 tps | 1.0M | $0.660 / $1.98 |
| Healthy | 100%◆ (best) | 96.47% | 95.97% | 95.97% | 788ms | 46.9s | 65 tps | 1.0M | $1.32 / $3.96 | |
| Healthy | 100%◆ (best) | 99.56% | 99.50% | 99.50% | 749ms◆ (best) | 12.3s | 98 tps | 1.0M | $1.32 / $3.96 | |
Togethertogether | Healthy | 100%◆ (best) | 97.37% | 97.17% | 97.17% | 1.15s | 18.7s | 77 tps | 1.0M | $1.32 / $3.96 |
| Healthy | 99.28% | 99.59% | 99.67%◆ (best) | 99.67%◆ (best) | 5.10s | 31.4s | 35 tps | 1.0M | $1.19 / $3.56 | |
| Healthy | 99.06% | 99.62% | 99.03% | 99.03% | 1.37s | 13.0s | 49 tps | 1.0M | $1.32 / $3.96 | |
| Degraded | 94.44% | 98.59% | 99.12% | 99.12% | 1.96s | 31.1s | 50 tps | 1.0M | $1.32 / $3.96 | |
DigitalOceandigitalocean | No data | — | 97.81% | 97.04% | 97.04% | 1.23s | 14.6s | 51 tps | 1.0M | $1.32 / $3.96 |
Fireworksfireworks | No data | — | 98.42% | 99.23% | 99.23% | 1.37s | 6.44s◆ (best) | 47 tps | 1.0M | $1.32 / $3.96 |
StreamLakestreamlake | No data | — | 98.21% | 98.72% | 98.72% | 3.39s | 33.3s | 33 tps | 1.0M | $1.32 / $3.96 |
- Alibaba100%◆ (best)Healthy24h 99.82%
- 7d
- 99.62%
- 30d
- 99.62%
- p50
- 1.29s
- p99
- 7.62s
- Throughput
- 54 tps
- Context
- 1.0M
- $/Mtok
- $0.603 / $1.81◆ (best)
alibaba - BaseTen100%◆ (best)Healthy24h 98.93%
- 7d
- 98.67%
- 30d
- 98.67%
- p50
- 1.54s
- p99
- 16.9s
- Throughput
- 103 tps◆ (best)
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
baseten/fp4fp4 - Cloudflare100%◆ (best)Healthy24h 98.09%
- 7d
- 97.21%
- 30d
- 97.21%
- p50
- 4.62s
- p99
- 47.1s
- Throughput
- 37 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
cloudflare - DeepSeek100%◆ (best)Healthy24h 99.95%
- 7d
- 86.81%
- 30d
- 86.81%
- p50
- 1.64s
- p99
- 9.50s
- Throughput
- 31 tps
- Context
- 1.0M
- $/Mtok
- $0.660 / $1.98
deepseek - Parasail100%◆ (best)Healthy24h 96.47%
- 7d
- 95.97%
- 30d
- 95.97%
- p50
- 788ms
- p99
- 46.9s
- Throughput
- 65 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
parasail/fp8fp8 - Sail Research100%◆ (best)Healthy24h 99.56%
- 7d
- 99.50%
- 30d
- 99.50%
- p50
- 749ms◆ (best)
- p99
- 12.3s
- Throughput
- 98 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
sail-research/fp4fp4 - Together100%◆ (best)Healthy24h 97.37%
- 7d
- 97.17%
- 30d
- 97.17%
- p50
- 1.15s
- p99
- 18.7s
- Throughput
- 77 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
together - GMICloud99.28%Healthy24h 99.59%
- 7d
- 99.67%◆ (best)
- 30d
- 99.67%◆ (best)
- p50
- 5.10s
- p99
- 31.4s
- Throughput
- 35 tps
- Context
- 1.0M
- $/Mtok
- $1.19 / $3.56
gmicloud/fp8fp8 - Novita99.06%Healthy24h 99.62%
- 7d
- 99.03%
- 30d
- 99.03%
- p50
- 1.37s
- p99
- 13.0s
- Throughput
- 49 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
novita/fp8fp8 - SiliconFlow94.44%Degraded24h 98.59%
- 7d
- 99.12%
- 30d
- 99.12%
- p50
- 1.96s
- p99
- 31.1s
- Throughput
- 50 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
siliconflow/fp8fp8 - No data24h 97.81%
- 7d
- 97.04%
- 30d
- 97.04%
- p50
- 1.23s
- p99
- 14.6s
- Throughput
- 51 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
digitalocean - No data24h 98.42%
- 7d
- 99.23%
- 30d
- 99.23%
- p50
- 1.37s
- p99
- 6.44s◆ (best)
- Throughput
- 47 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
fireworks - No data24h 98.21%
- 7d
- 98.72%
- 30d
- 98.72%
- p50
- 3.39s
- p99
- 33.3s
- Throughput
- 33 tps
- Context
- 1.0M
- $/Mtok
- $1.32 / $3.96
streamlake
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.
Throughput
Median tokens per second.