DeepSeek V4 Flash 0423

deepseek/deepseek-v4-flash

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

CompareDeepSeek V4 Flash 0423

	State
Cloudflarecloudflare	Healthy	100%	99.99%	816ms	10.2s	55 tps	384K	$0.440 / $1.32
DeepSeekdeepseek	Healthy	100%	99.89%	837ms	4.82s	74 tps	1.0M	$0.220 / $0.660
Novitanovita/fp8fp8	Healthy	99.97%	99.85%	1.07s	5.22s	73 tps	1.0M	$0.140 / $0.280
Baidubaidu/fp8fp8	Healthy	99.96%	99.16%	667ms	5.12s	87 tps	1.0M	$0.080 / $0.160
GMICloudgmicloud/fp8fp8	Healthy	99.94%	99.84%	2.21s	6.06s	44 tps	1.0M	$0.112 / $0.224
AtlasCloudatlas-cloud/fp4fp4	Healthy	99.94%	99.91%	997ms	4.20s	46 tps	1.0M	$0.140 / $0.280
SiliconFlowsiliconflow/fp8fp8	Healthy	99.93%	99.37%	1.37s	9.82s	46 tps	1.0M	$0.130 / $0.280
StreamLakestreamlake/fp8fp8	Healthy	99.91%	99.30%	1.53s	11.4s	46 tps	1.0M	$0.081 / $0.162
Venicevenice	Healthy	99.90%	99.77%	1.61s	4.60s	35 tps	1.0M	$0.138 / $0.275
Alibabaalibaba/fp8fp8	Healthy	99.90%	99.82%	838ms	3.61s	87 tps	1.0M	$0.134 / $0.268
Sail Researchsail-research/fp4fp4	Healthy	99.82%	99.72%	761ms	5.88s	20 tps	1.0M	$0.090 / $0.180
DeepInfradeepinfra/fp8fp8	Healthy	99.77%	99.59%	746ms	2.72s	30 tps	1.0M	$0.090 / $0.180
CoreWeavecoreweave/fp8fp8	Healthy	99.62%	99.27%	527ms	10.6s	40 tps	1.0M	$0.140 / $0.280
Mancer 2mancer/fp8fp8	Healthy	99.48%	98.65%	1.00s	8.21s	33 tps	1.0M	$0.145 / $0.450
Parasailparasail/fp8fp8	Healthy	99.24%	99.16%	659ms	10.1s	53 tps	1.0M	$0.140 / $0.280
DigitalOceandigitalocean	Healthy	98.21%	99.07%	1.81s	53.7s	5.0 tps	1.0M	$0.068 / $0.168
Azureazure/us	Down	94.52%	84.78%	2.61s	29.7s	19 tps	1.0M	$0.210 / $0.560
Phalaphala	Down	—	94.10%	—	—	—	1.0M	$0.200 / $0.400

Cloudflare100%
Healthy24h 99.99%
p50
816ms
p99
10.2s
Throughput
55 tps
Context
384K
$/Mtok
$0.440 / $1.32
cloudflare
DeepSeek100%
Healthy24h 99.89%
p50
837ms
p99
4.82s
Throughput
74 tps
Context
1.0M
$/Mtok
$0.220 / $0.660
deepseek
Novita99.97%
Healthy24h 99.85%
p50
1.07s
p99
5.22s
Throughput
73 tps
Context
1.0M
$/Mtok
$0.140 / $0.280
novita/fp8fp8
Baidu99.96%
Healthy24h 99.16%
p50
667ms
p99
5.12s
Throughput
87 tps
Context
1.0M
$/Mtok
$0.080 / $0.160
baidu/fp8fp8
GMICloud99.94%
Healthy24h 99.84%
p50
2.21s
p99
6.06s
Throughput
44 tps
Context
1.0M
$/Mtok
$0.112 / $0.224
gmicloud/fp8fp8
AtlasCloud99.94%
Healthy24h 99.91%
p50
997ms
p99
4.20s
Throughput
46 tps
Context
1.0M
$/Mtok
$0.140 / $0.280
atlas-cloud/fp4fp4
SiliconFlow99.93%
Healthy24h 99.37%
p50
1.37s
p99
9.82s
Throughput
46 tps
Context
1.0M
$/Mtok
$0.130 / $0.280
siliconflow/fp8fp8
StreamLake99.91%
Healthy24h 99.30%
p50
1.53s
p99
11.4s
Throughput
46 tps
Context
1.0M
$/Mtok
$0.081 / $0.162
streamlake/fp8fp8
Venice99.90%
Healthy24h 99.77%
p50
1.61s
p99
4.60s
Throughput
35 tps
Context
1.0M
$/Mtok
$0.138 / $0.275
venice
Alibaba99.90%
Healthy24h 99.82%
p50
838ms
p99
3.61s
Throughput
87 tps
Context
1.0M
$/Mtok
$0.134 / $0.268
alibaba/fp8fp8
Sail Research99.82%
Healthy24h 99.72%
p50
761ms
p99
5.88s
Throughput
20 tps
Context
1.0M
$/Mtok
$0.090 / $0.180
sail-research/fp4fp4
DeepInfra99.77%
Healthy24h 99.59%
p50
746ms
p99
2.72s
Throughput
30 tps
Context
1.0M
$/Mtok
$0.090 / $0.180
deepinfra/fp8fp8
CoreWeave99.62%
Healthy24h 99.27%
p50
527ms
p99
10.6s
Throughput
40 tps
Context
1.0M
$/Mtok
$0.140 / $0.280
coreweave/fp8fp8
Mancer 299.48%
Healthy24h 98.65%
p50
1.00s
p99
8.21s
Throughput
33 tps
Context
1.0M
$/Mtok
$0.145 / $0.450
mancer/fp8fp8
Parasail99.24%
Healthy24h 99.16%
p50
659ms
p99
10.1s
Throughput
53 tps
Context
1.0M
$/Mtok
$0.140 / $0.280
parasail/fp8fp8
DigitalOcean98.21%
Healthy24h 99.07%
p50
1.81s
p99
53.7s
Throughput
5.0 tps
Context
1.0M
$/Mtok
$0.068 / $0.168
digitalocean
Azure94.52%
Down24h 84.78%
p50
2.61s
p99
29.7s
Throughput
19 tps
Context
1.0M
$/Mtok
$0.210 / $0.560
azure/us
Phala—
Down24h 94.10%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$0.200 / $0.400
phala

Measured 4m 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 is OpenRouter's rolling 30-minute window, averaged over the hour.

Throughput

Median tokens per second.