DeepSeek V3 0324: 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) | 99.44% | 98.98% | 98.98% | 2.78s | 6.77s | 17 tps | 164K | $0.290 / $1.14 | |
| Healthy | 99.47% | 99.94% | 99.94%◆ (best) | 99.94%◆ (best) | 295ms◆ (best) | 5.07s◆ (best) | 32 tps◆ (best) | 164K | $0.500 / $1.50 | |
| Healthy | 97.91% | 98.59% | 97.29% | 97.29% | 1.46s | 10.9s | 18 tps | 164K | $0.250 / $1.00 | |
| Healthy | 96.92% | 98.29% | 97.79% | 97.79% | 1.22s | 50.1s | 29 tps | 164K | $0.270 / $1.12 | |
| Degraded | 96.77% | 91.73% | 78.82% | 78.82% | 948ms | 97.7s | 23 tps | 164K | $0.240 / $0.900◆ (best) |
- GMICloud100%◆ (best)Healthy24h 99.44%
- 7d
- 98.98%
- 30d
- 98.98%
- p50
- 2.78s
- p99
- 6.77s
- Throughput
- 17 tps
- Context
- 164K
- $/Mtok
- $0.290 / $1.14
gmicloud/fp8fp8 - Crusoe99.47%Healthy24h 99.94%
- 7d
- 99.94%◆ (best)
- 30d
- 99.94%◆ (best)
- p50
- 295ms◆ (best)
- p99
- 5.07s◆ (best)
- Throughput
- 32 tps◆ (best)
- Context
- 164K
- $/Mtok
- $0.500 / $1.50
crusoe/bf16bf16 - SiliconFlow97.91%Healthy24h 98.59%
- 7d
- 97.29%
- 30d
- 97.29%
- p50
- 1.46s
- p99
- 10.9s
- Throughput
- 18 tps
- Context
- 164K
- $/Mtok
- $0.250 / $1.00
siliconflow/fp8fp8 - Novita96.92%Healthy24h 98.29%
- 7d
- 97.79%
- 30d
- 97.79%
- p50
- 1.22s
- p99
- 50.1s
- Throughput
- 29 tps
- Context
- 164K
- $/Mtok
- $0.270 / $1.12
novita/fp8fp8 - DeepInfra96.77%Degraded24h 91.73%
- 7d
- 78.82%
- 30d
- 78.82%
- p50
- 948ms
- p99
- 97.7s
- Throughput
- 23 tps
- Context
- 164K
- $/Mtok
- $0.240 / $0.900◆ (best)
deepinfra/fp4fp4
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.