DeepSeek V3.1: 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) | 100% | 99.99%◆ (best) | 99.99%◆ (best) | 365ms◆ (best) | 944ms◆ (best) | 45 tps | 161K | $0.550 / $1.65 | |
| Healthy | 100%◆ (best) | 99.85% | 99.53% | 99.53% | 805ms | 8.39s | 11 tps | 164K | $0.250 / $0.950◆ (best) | |
Googlegoogle-vertex/us-west2 | Healthy | 100%◆ (best) | 97.71% | 95.72% | 95.72% | 1.16s | 2.64s | 75 tps | 164K | $0.600 / $1.70 |
| Healthy | 100%◆ (best) | 99.99% | 99.99% | 99.99% | 1.66s | 9.05s | 21 tps | 131K | $0.270 / $1.00 | |
| Degraded | 100%◆ (best) | 97.88% | 98.03% | 98.03% | 1.54s | 13.9s | 15 tps | 164K | $0.270 / $1.00 | |
| Healthy | 99.81% | 99.91% | 99.87% | 99.87% | 1.69s | 4.61s | 27 tps | 131K | $0.300 / $0.950 | |
Maramara | Down | 95.24% | 94.38% | 82.90% | 82.90% | 725ms | 11.2s | 92 tps◆ (best) | 131K | $0.600 / $1.70 |
| No data | — | 97.33% | 97.61% | 97.61% | 2.19s | 12.5s | 25 tps | 131K | $0.650 / $1.50 |
- CoreWeave100%◆ (best)Healthy24h 100%
- 7d
- 99.99%◆ (best)
- 30d
- 99.99%◆ (best)
- p50
- 365ms◆ (best)
- p99
- 944ms◆ (best)
- Throughput
- 45 tps
- Context
- 161K
- $/Mtok
- $0.550 / $1.65
coreweave/fp8fp8 - DeepInfra100%◆ (best)Healthy24h 99.85%
- 7d
- 99.53%
- 30d
- 99.53%
- p50
- 805ms
- p99
- 8.39s
- Throughput
- 11 tps
- Context
- 164K
- $/Mtok
- $0.250 / $0.950◆ (best)
deepinfra/fp4fp4 - Google100%◆ (best)Healthy24h 97.71%
- 7d
- 95.72%
- 30d
- 95.72%
- p50
- 1.16s
- p99
- 2.64s
- Throughput
- 75 tps
- Context
- 164K
- $/Mtok
- $0.600 / $1.70
google-vertex/us-west2 - Novita100%◆ (best)Healthy24h 99.99%
- 7d
- 99.99%
- 30d
- 99.99%
- p50
- 1.66s
- p99
- 9.05s
- Throughput
- 21 tps
- Context
- 131K
- $/Mtok
- $0.270 / $1.00
novita/fp8fp8 - SiliconFlow100%◆ (best)Degraded24h 97.88%
- 7d
- 98.03%
- 30d
- 98.03%
- p50
- 1.54s
- p99
- 13.9s
- Throughput
- 15 tps
- Context
- 164K
- $/Mtok
- $0.270 / $1.00
siliconflow/fp8fp8 - AtlasCloud99.81%Healthy24h 99.91%
- 7d
- 99.87%
- 30d
- 99.87%
- p50
- 1.69s
- p99
- 4.61s
- Throughput
- 27 tps
- Context
- 131K
- $/Mtok
- $0.300 / $0.950
atlas-cloud/fp8fp8 - Mara95.24%Down24h 94.38%
- 7d
- 82.90%
- 30d
- 82.90%
- p50
- 725ms
- p99
- 11.2s
- Throughput
- 92 tps◆ (best)
- Context
- 131K
- $/Mtok
- $0.600 / $1.70
mara - No data24h 97.33%
- 7d
- 97.61%
- 30d
- 97.61%
- p50
- 2.19s
- p99
- 12.5s
- Throughput
- 25 tps
- Context
- 131K
- $/Mtok
- $0.650 / $1.50
sambanova/fp8fp8
Measured 4m 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.