DeepSeek V3
deepseek/deepseek-chat
Compare with DeepSeek V4 Flash 0731, DeepSeek V4 Flash 0423, DeepSeek V4 Pro 0423
CompareDeepSeek V3
| Provider | State | Uptime 5m | 24h | p50 latency | p99 | Throughput | Context | $/Mtok |
|---|---|---|---|---|---|---|---|---|
Novitanovita/fp8fp8 | Healthy | 100% | 99.93% | 1.11s | 3.80s | 22 tps | 64K | $0.400 / $1.30 |
DeepInfradeepinfra/fp4fp4 | Degraded | 98.79% | 95.76% | 839ms | 7.50s | 16 tps | 164K | $0.320 / $0.890 |
StreamLakestreamlake | Healthy | 98.47% | 99.47% | 1.42s | 6.00s | 33 tps | 128K | $0.257 / $1.03 |
- Novita100%Healthy24h 99.93%
- p50
- 1.11s
- p99
- 3.80s
- Throughput
- 22 tps
- Context
- 64K
- $/Mtok
- $0.400 / $1.30
novita/fp8fp8 - DeepInfra98.79%Degraded24h 95.76%
- p50
- 839ms
- p99
- 7.50s
- Throughput
- 16 tps
- Context
- 164K
- $/Mtok
- $0.320 / $0.890
deepinfra/fp4fp4 - StreamLake98.47%Healthy24h 99.47%
- p50
- 1.42s
- p99
- 6.00s
- Throughput
- 33 tps
- Context
- 128K
- $/Mtok
- $0.257 / $1.03
streamlake
Measured 3m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.
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 and p99 are OpenRouter's rolling 30-minute windows, averaged over the hour.
Throughput
Median tokens per second.