DeepSeek V4 Pro 0423

deepseek/deepseek-v4-pro

Compare with DeepSeek V4 Pro 0813, DeepSeek V4 Flash 0731, DeepSeek V4 Flash 0423

CompareDeepSeek V4 Pro 0423

Provider	State	Uptime 5m	24h	p50 latency	p99	Throughput	Context	$/Mtok
BaseTenbaseten/fp4fp4	Healthy	100%	99.13%	427ms	2.95s	63 tps	1.0M	$1.74 / $3.48
DeepSeekdeepseek	Healthy	100%	77.29%	1.15s	3.85s	34 tps	1.0M	$0.660 / $1.98
Novitanovita/fp8fp8	Healthy	99.94%	99.12%	1.50s	12.3s	60 tps	1.0M	$1.44 / $2.88
Baidubaidu/fp8fp8	Healthy	99.80%	97.65%	834ms	7.03s	66 tps	1.0M	$1.69 / $3.38
Alibabaalibaba/fp8fp8	Healthy	99.55%	97.56%	1.42s	4.89s	51 tps	1.0M	$1.42 / $2.83
SiliconFlowsiliconflow/fp8fp8	Degraded	99.41%	96.00%	1.50s	100s	39 tps	1.0M	$1.50 / $3.14
GMICloudgmicloud/fp8fp8	Healthy	99.23%	95.70%	3.10s	10.4s	27 tps	1.0M	$0.696 / $1.39
AtlasCloudatlas-cloud/fp4fp4	Degraded	99.00%	92.58%	1.42s	7.76s	37 tps	1.0M	$1.68 / $3.38
StreamLakestreamlake/fp8fp8	Healthy	98.99%	96.12%	2.17s	18.4s	39 tps	1.0M	$0.694 / $1.39
Parasailparasail/fp8fp8	Degraded	98.81%	89.69%	737ms	7.03s	49 tps	1.0M	$1.74 / $3.48
Ionstreamionstream/fp4fp4	Degraded	98.24%	95.42%	2.71s	13.1s	23 tps	1.0M	$1.13 / $2.26
Togethertogether	Healthy	97.05%	90.50%	638ms	12.4s	39 tps	512K	$1.74 / $3.48
Azureazure/us	Degraded	96.89%	93.60%	1.40s	10.6s	41 tps	1.0M	$1.91 / $3.83
Venicevenice	Down	54.29%	74.22%	3.04s	45.9s	39 tps	1.0M	$1.65 / $3.30
CoreWeavecoreweave/fp8fp8	No data	—	97.31%	652ms	13.3s	62 tps	1.0M	$1.15 / $2.55
DeepInfradeepinfra/fp8fp8	No data	—	96.32%	1.92s	35.0s	8.0 tps	1.0M	$1.30 / $2.60
DigitalOceandigitalocean	Down	—	93.91%	—	—	—	1.0M	$0.870 / $1.74
Fireworksfireworks	No data	—	0.00%	2.22s	10.6s	46 tps	1.0M	$1.74 / $3.48

BaseTen100%
Healthy24h 99.13%
p50
427ms
p99
2.95s
Throughput
63 tps
Context
1.0M
$/Mtok
$1.74 / $3.48
baseten/fp4fp4
DeepSeek100%
Healthy24h 77.29%
p50
1.15s
p99
3.85s
Throughput
34 tps
Context
1.0M
$/Mtok
$0.660 / $1.98
deepseek
Novita99.94%
Healthy24h 99.12%
p50
1.50s
p99
12.3s
Throughput
60 tps
Context
1.0M
$/Mtok
$1.44 / $2.88
novita/fp8fp8
Baidu99.80%
Healthy24h 97.65%
p50
834ms
p99
7.03s
Throughput
66 tps
Context
1.0M
$/Mtok
$1.69 / $3.38
baidu/fp8fp8
Alibaba99.55%
Healthy24h 97.56%
p50
1.42s
p99
4.89s
Throughput
51 tps
Context
1.0M
$/Mtok
$1.42 / $2.83
alibaba/fp8fp8
SiliconFlow99.41%
Degraded24h 96.00%
p50
1.50s
p99
100s
Throughput
39 tps
Context
1.0M
$/Mtok
$1.50 / $3.14
siliconflow/fp8fp8
GMICloud99.23%
Healthy24h 95.70%
p50
3.10s
p99
10.4s
Throughput
27 tps
Context
1.0M
$/Mtok
$0.696 / $1.39
gmicloud/fp8fp8
AtlasCloud99.00%
Degraded24h 92.58%
p50
1.42s
p99
7.76s
Throughput
37 tps
Context
1.0M
$/Mtok
$1.68 / $3.38
atlas-cloud/fp4fp4
StreamLake98.99%
Healthy24h 96.12%
p50
2.17s
p99
18.4s
Throughput
39 tps
Context
1.0M
$/Mtok
$0.694 / $1.39
streamlake/fp8fp8
Parasail98.81%
Degraded24h 89.69%
p50
737ms
p99
7.03s
Throughput
49 tps
Context
1.0M
$/Mtok
$1.74 / $3.48
parasail/fp8fp8
Ionstream98.24%
Degraded24h 95.42%
p50
2.71s
p99
13.1s
Throughput
23 tps
Context
1.0M
$/Mtok
$1.13 / $2.26
ionstream/fp4fp4
Together97.05%
Healthy24h 90.50%
p50
638ms
p99
12.4s
Throughput
39 tps
Context
512K
$/Mtok
$1.74 / $3.48
together
Azure96.89%
Degraded24h 93.60%
p50
1.40s
p99
10.6s
Throughput
41 tps
Context
1.0M
$/Mtok
$1.91 / $3.83
azure/us
Venice54.29%
Down24h 74.22%
p50
3.04s
p99
45.9s
Throughput
39 tps
Context
1.0M
$/Mtok
$1.65 / $3.30
venice
CoreWeave—
No data24h 97.31%
p50
652ms
p99
13.3s
Throughput
62 tps
Context
1.0M
$/Mtok
$1.15 / $2.55
coreweave/fp8fp8
DeepInfra—
No data24h 96.32%
p50
1.92s
p99
35.0s
Throughput
8.0 tps
Context
1.0M
$/Mtok
$1.30 / $2.60
deepinfra/fp8fp8
DigitalOcean—
Down24h 93.91%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$0.870 / $1.74
digitalocean
Fireworks—
No data24h 0.00%
p50
2.22s
p99
10.6s
Throughput
46 tps
Context
1.0M
$/Mtok
$1.74 / $3.48
fireworks

Measured 4m ago. Latency and throughput are OpenRouter's rolling 30-minute windows.

History

3 endpoints reporting no data are omitted from the charts.

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.