DeepSeek V3.2: 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
Baidubaidu/fp8fp8	Healthy	100%◆ (best)	99.91%	99.94%	99.94%	1.01s	3.96s◆ (best)	51 tps	131K	$0.280 / $0.420
DigitalOceandigitalocean	Healthy	100%◆ (best)	96.48%	97.18%	97.18%	796ms	7.32s	42 tps	164K	$0.250 / $0.800
Friendlifriendli	Healthy	100%◆ (best)	99.88%	91.94%	91.94%	373ms◆ (best)	6.29s	44 tps	164K	$0.500 / $1.50
GMICloudgmicloud/fp8fp8	Healthy	100%◆ (best)	99.54%	99.56%	99.56%	1.97s	7.82s	9.0 tps	164K	$0.209 / $0.310◆ (best)
Googlegoogle-vertex	Healthy	100%◆ (best)	99.44%	99.69%	99.69%	964ms	10.9s	20 tps	164K	$0.560 / $1.68
Novitanovita/fp8fp8	Healthy	100%◆ (best)	99.98%	99.99%◆ (best)	99.99%◆ (best)	1.28s	6.06s	25 tps	164K	$0.269 / $0.400
AtlasCloudatlas-cloud/fp8fp8	Healthy	99.89%	99.96%	99.96%	99.96%	1.51s	12.7s	21 tps	164K	$0.260 / $0.380
Alibabaalibaba/fp8fp8	Healthy	99.57%	97.94%	98.39%	98.39%	783ms	11.1s	52 tps◆ (best)	131K	$0.370 / $1.11
StreamLakestreamlake/fp8fp8	Healthy	99.08%	97.47%	99.41%	99.41%	1.43s	8.54s	23 tps	128K	$0.214 / $0.322
SiliconFlowsiliconflow/fp8fp8	Healthy	96.79%	95.88%	99.20%	99.20%	1.82s	9.93s	25 tps	164K	$0.259 / $0.420
DeepInfradeepinfra/fp4fp4	No data	—	97.21%	93.97%	93.97%	2.41s	41.8s	7.0 tps	164K	$0.260 / $0.380
Phalaphala	Degraded	—	98.28%	99.06%	99.06%	4.68s	335s	5.0 tps	164K	$1.00 / $1.00
SambaNovasambanova	No data	—	95.63%	96.95%	96.95%	—	—	—	33K	$3.00 / $4.50
Venicevenice	Down	—	82.93%	75.92%	75.92%	3.65s	59.1s	5.0 tps	160K	$0.330 / $0.480

Baidu100%◆ (best)
Healthy24h 99.91%
7d
99.94%
30d
99.94%
p50
1.01s
p99
3.96s◆ (best)
Throughput
51 tps
Context
131K
$/Mtok
$0.280 / $0.420
baidu/fp8fp8
DigitalOcean100%◆ (best)
Healthy24h 96.48%
7d
97.18%
30d
97.18%
p50
796ms
p99
7.32s
Throughput
42 tps
Context
164K
$/Mtok
$0.250 / $0.800
digitalocean
Friendli100%◆ (best)
Healthy24h 99.88%
7d
91.94%
30d
91.94%
p50
373ms◆ (best)
p99
6.29s
Throughput
44 tps
Context
164K
$/Mtok
$0.500 / $1.50
friendli
GMICloud100%◆ (best)
Healthy24h 99.54%
7d
99.56%
30d
99.56%
p50
1.97s
p99
7.82s
Throughput
9.0 tps
Context
164K
$/Mtok
$0.209 / $0.310◆ (best)
gmicloud/fp8fp8
Google100%◆ (best)
Healthy24h 99.44%
7d
99.69%
30d
99.69%
p50
964ms
p99
10.9s
Throughput
20 tps
Context
164K
$/Mtok
$0.560 / $1.68
google-vertex
Novita100%◆ (best)
Healthy24h 99.98%
7d
99.99%◆ (best)
30d
99.99%◆ (best)
p50
1.28s
p99
6.06s
Throughput
25 tps
Context
164K
$/Mtok
$0.269 / $0.400
novita/fp8fp8
AtlasCloud99.89%
Healthy24h 99.96%
7d
99.96%
30d
99.96%
p50
1.51s
p99
12.7s
Throughput
21 tps
Context
164K
$/Mtok
$0.260 / $0.380
atlas-cloud/fp8fp8
Alibaba99.57%
Healthy24h 97.94%
7d
98.39%
30d
98.39%
p50
783ms
p99
11.1s
Throughput
52 tps◆ (best)
Context
131K
$/Mtok
$0.370 / $1.11
alibaba/fp8fp8
StreamLake99.08%
Healthy24h 97.47%
7d
99.41%
30d
99.41%
p50
1.43s
p99
8.54s
Throughput
23 tps
Context
128K
$/Mtok
$0.214 / $0.322
streamlake/fp8fp8
SiliconFlow96.79%
Healthy24h 95.88%
7d
99.20%
30d
99.20%
p50
1.82s
p99
9.93s
Throughput
25 tps
Context
164K
$/Mtok
$0.259 / $0.420
siliconflow/fp8fp8
DeepInfra—
No data24h 97.21%
7d
93.97%
30d
93.97%
p50
2.41s
p99
41.8s
Throughput
7.0 tps
Context
164K
$/Mtok
$0.260 / $0.380
deepinfra/fp4fp4
Phala—
Degraded24h 98.28%
7d
99.06%
30d
99.06%
p50
4.68s
p99
335s
Throughput
5.0 tps
Context
164K
$/Mtok
$1.00 / $1.00
phala
SambaNova—
No data24h 95.63%
7d
96.95%
30d
96.95%
p50
—
p99
—
Throughput
—
Context
33K
$/Mtok
$3.00 / $4.50
sambanova
Venice—
Down24h 82.93%
7d
75.92%
30d
75.92%
p50
3.65s
p99
59.1s
Throughput
5.0 tps
Context
160K
$/Mtok
$0.330 / $0.480
venice

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.