GLM 5.3: 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
AkashMLakashml/fp8fp8	Healthy	100%◆ (best)	99.77%	99.36%	97.76%	773ms	3.16s	78 tps	1.0M	$1.30 / $4.40
Alibabaalibaba	Healthy	100%◆ (best)	99.97%	98.35%	98.35%	917ms	5.08s	79 tps	1.0M	$1.40 / $4.40
AtlasCloudatlas-cloud/fp8fp8	Healthy	100%◆ (best)	99.90%	99.92%	99.62%	3.32s	8.41s	59 tps	1.0M	$1.40 / $4.40
BaseTenbaseten/fp4fp4	Healthy	100%◆ (best)	99.75%	97.93%	98.53%	1.46s	11.3s	62 tps	1.0M	$1.40 / $4.40
Crusoecrusoe/fp4fp4	Healthy	100%◆ (best)	97.38%	94.09%	95.69%	633ms	12.6s	103 tps	1.0M	$1.40 / $4.40
DigitalOceandigitalocean	Healthy	100%◆ (best)	98.96%	95.83%	97.53%	940ms	4.49s	46 tps	1.0M	$1.08 / $3.39
Fireworksfireworks	Healthy	100%◆ (best)	99.60%	99.73%	99.56%	1.13s	39.9s	46 tps	1.0M	$1.40 / $4.40
Friendlifriendli	Healthy	100%◆ (best)	99.93%	99.98%	99.92%	714ms	7.69s	106 tps	1.0M	$1.26 / $3.96
Inceptroninceptron/fp4fp4	Healthy	100%◆ (best)	98.49%	98.52%	98.16%	1.16s	14.1s	77 tps	1.0M	$1.07 / $4.09
InferenceNetinference-net/fp4fp4	Healthy	100%◆ (best)	91.94%	52.67%	52.67%	871ms	48.2s	58 tps	1.0M	$0.900 / $3.00
Io Netio-net/fp8fp8	Healthy	100%◆ (best)	99.53%	99.64%	98.98%	1.82s	19.3s	55 tps	262K	$0.910 / $3.08
Morphmorph/fp8fp8	Healthy	100%◆ (best)	99.40%	98.63%	98.63%	1.54s	11.7s	67 tps	1.0M	$1.01 / $3.18
Parasailparasail/fp8fp8	Healthy	100%◆ (best)	99.19%	99.03%	97.09%	910ms	57.2s	95 tps	1.0M	$1.40 / $4.40
Phalaphala	Healthy	100%◆ (best)	99.71%	99.03%	98.95%	1.08s	16.1s	53 tps	1.0M	$1.05 / $3.30
Sail Researchsail-research/usfp8	Healthy	100%◆ (best)	99.89%	99.91%	99.91%	1.12s	9.40s	65 tps	1.0M	$1.00 / $3.16
Z.AIz-ai/fp8fp8	Healthy	100%◆ (best)	99.94%	99.78%	99.91%	3.10s	30.6s	48 tps	1.0M	$1.40 / $4.40
Modalmodal	Healthy	99.94%	98.13%	99.28%	99.55%	865ms	34.5s	71 tps	1.0M	$1.40 / $4.40
Novitanovita/fp8fp8	Healthy	99.63%	97.52%	99.43%	99.62%	3.82s	60.6s	32 tps	1.0M	$0.910 / $2.86
Waferwafer	Healthy	99.60%	99.46%	98.31%	97.45%	517ms	14.5s	107 tps	1.0M	$1.40 / $4.40
Rekareka/fp8fp8	Healthy	99.30%	98.95%	99.26%	97.14%	2.21s	41.4s	19 tps	262K	$0.877 / $2.97
DeepInfradeepinfra/fp4fp4	Degraded	99.13%	95.49%	97.04%	96.24%	3.78s	67.5s	21 tps	1.0M	$0.900 / $3.00
Sail Researchsail-research/fp8fp8	Healthy	98.31%	99.92%	99.93%	99.94%	1.25s	4.24s	69 tps	1.0M	$1.00 / $3.16
Makoramakora/fp4fp4	Healthy	98.17%	95.87%	96.83%	94.23%	933ms	54.1s	83 tps	980K	$1.05 / $4.20
Togethertogether	Down	84.94%	94.67%	96.78%	96.76%	329ms◆ (best)	40.2s	112 tps◆ (best)	1.0M	$1.40 / $4.40
Decartdecart/fp4fp4	Down	2.13%	98.19%	99.30%	97.95%	1.22s	21.2s	95 tps	1.0M	$1.19 / $3.74
Baidubaidu/fp8fp8	No data	—	100%	100%◆ (best)	100%◆ (best)	783ms	1.30s◆ (best)	77 tps	1.0M	$1.40 / $4.40
BaseTenbaseten/fp8fp8	No data	—	95.88%	99.21%	99.59%	1.23s	34.8s	62 tps	1.0M	$2.10 / $6.60
Cloudflarecloudflare	No data	—	99.95%	99.78%	99.72%	2.77s	68.0s	35 tps	1.3M	$1.40 / $4.40
Crusoecrusoe/fp8fp8	No data	—	—	—	99.41%	—	—	—	1.0M	$1.40 / $4.40
DeepInfradeepinfra/bf16bf16	No data	—	—	—	88.65%	—	—	—	1.0M	$1.20 / $4.00
DeepInfradeepinfra/fp8fp8	No data	—	—	—	95.08%	—	—	—	1.0M	$1.20 / $4.00
Fireworksfireworks/us	No data	—	—	—	—	—	—	—	1.0M	$2.10 / $6.60
GMICloudgmicloud/fp8fp8	No data	—	98.53%	99.26%	98.74%	2.62s	14.4s	43 tps	1.0M	$1.05 / $3.30
Io Netio-net/fp4fp4	No data	—	—	100%◆ (best)	99.72%	—	—	—	262K	$1.13 / $4.32
Mistralmistral/nvfp4nvfp4	No data	—	98.84%	99.34%	99.34%	4.03s	209s	46 tps	1.0M	$1.40 / $4.40
Morphmorph/fp4fp4	No data	—	—	100%◆ (best)	95.37%	—	—	—	1.0M	$1.00 / $3.41
SiliconFlowsiliconflow/fp8fp8	No data	—	99.60%	99.22%	99.29%	2.00s	14.7s	47 tps	1.0M	$1.12 / $3.52
StreamLakestreamlake	No data	—	—	—	100%◆ (best)	—	—	—	1.0M	$1.11 / $3.50
StreamLakestreamlake/fp8fp8	No data	—	—	99.65%	98.85%	—	—	—	1.0M	$0.700 / $2.20◆ (best)
Venicevenice	No data	—	97.42%	98.51%	96.21%	695ms	4.40s	48 tps	1.0M	$1.40 / $4.40

AkashML100%◆ (best)
Healthy24h 99.77%
7d
99.36%
30d
97.76%
p50
773ms
p99
3.16s
Throughput
78 tps
Context
1.0M
$/Mtok
$1.30 / $4.40
akashml/fp8fp8
Alibaba100%◆ (best)
Healthy24h 99.97%
7d
98.35%
30d
98.35%
p50
917ms
p99
5.08s
Throughput
79 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
alibaba
AtlasCloud100%◆ (best)
Healthy24h 99.90%
7d
99.92%
30d
99.62%
p50
3.32s
p99
8.41s
Throughput
59 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
atlas-cloud/fp8fp8
BaseTen100%◆ (best)
Healthy24h 99.75%
7d
97.93%
30d
98.53%
p50
1.46s
p99
11.3s
Throughput
62 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
baseten/fp4fp4
Crusoe100%◆ (best)
Healthy24h 97.38%
7d
94.09%
30d
95.69%
p50
633ms
p99
12.6s
Throughput
103 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
crusoe/fp4fp4
DigitalOcean100%◆ (best)
Healthy24h 98.96%
7d
95.83%
30d
97.53%
p50
940ms
p99
4.49s
Throughput
46 tps
Context
1.0M
$/Mtok
$1.08 / $3.39
digitalocean
Fireworks100%◆ (best)
Healthy24h 99.60%
7d
99.73%
30d
99.56%
p50
1.13s
p99
39.9s
Throughput
46 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
fireworks
Friendli100%◆ (best)
Healthy24h 99.93%
7d
99.98%
30d
99.92%
p50
714ms
p99
7.69s
Throughput
106 tps
Context
1.0M
$/Mtok
$1.26 / $3.96
friendli
Inceptron100%◆ (best)
Healthy24h 98.49%
7d
98.52%
30d
98.16%
p50
1.16s
p99
14.1s
Throughput
77 tps
Context
1.0M
$/Mtok
$1.07 / $4.09
inceptron/fp4fp4
InferenceNet100%◆ (best)
Healthy24h 91.94%
7d
52.67%
30d
52.67%
p50
871ms
p99
48.2s
Throughput
58 tps
Context
1.0M
$/Mtok
$0.900 / $3.00
inference-net/fp4fp4
Io Net100%◆ (best)
Healthy24h 99.53%
7d
99.64%
30d
98.98%
p50
1.82s
p99
19.3s
Throughput
55 tps
Context
262K
$/Mtok
$0.910 / $3.08
io-net/fp8fp8
Morph100%◆ (best)
Healthy24h 99.40%
7d
98.63%
30d
98.63%
p50
1.54s
p99
11.7s
Throughput
67 tps
Context
1.0M
$/Mtok
$1.01 / $3.18
morph/fp8fp8
Parasail100%◆ (best)
Healthy24h 99.19%
7d
99.03%
30d
97.09%
p50
910ms
p99
57.2s
Throughput
95 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
parasail/fp8fp8
Phala100%◆ (best)
Healthy24h 99.71%
7d
99.03%
30d
98.95%
p50
1.08s
p99
16.1s
Throughput
53 tps
Context
1.0M
$/Mtok
$1.05 / $3.30
phala
Sail Research100%◆ (best)
Healthy24h 99.89%
7d
99.91%
30d
99.91%
p50
1.12s
p99
9.40s
Throughput
65 tps
Context
1.0M
$/Mtok
$1.00 / $3.16
sail-research/usfp8
Z.AI100%◆ (best)
Healthy24h 99.94%
7d
99.78%
30d
99.91%
p50
3.10s
p99
30.6s
Throughput
48 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
z-ai/fp8fp8
Modal99.94%
Healthy24h 98.13%
7d
99.28%
30d
99.55%
p50
865ms
p99
34.5s
Throughput
71 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
modal
Novita99.63%
Healthy24h 97.52%
7d
99.43%
30d
99.62%
p50
3.82s
p99
60.6s
Throughput
32 tps
Context
1.0M
$/Mtok
$0.910 / $2.86
novita/fp8fp8
Wafer99.60%
Healthy24h 99.46%
7d
98.31%
30d
97.45%
p50
517ms
p99
14.5s
Throughput
107 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
wafer
Reka99.30%
Healthy24h 98.95%
7d
99.26%
30d
97.14%
p50
2.21s
p99
41.4s
Throughput
19 tps
Context
262K
$/Mtok
$0.877 / $2.97
reka/fp8fp8
DeepInfra99.13%
Degraded24h 95.49%
7d
97.04%
30d
96.24%
p50
3.78s
p99
67.5s
Throughput
21 tps
Context
1.0M
$/Mtok
$0.900 / $3.00
deepinfra/fp4fp4
Sail Research98.31%
Healthy24h 99.92%
7d
99.93%
30d
99.94%
p50
1.25s
p99
4.24s
Throughput
69 tps
Context
1.0M
$/Mtok
$1.00 / $3.16
sail-research/fp8fp8
Makora98.17%
Healthy24h 95.87%
7d
96.83%
30d
94.23%
p50
933ms
p99
54.1s
Throughput
83 tps
Context
980K
$/Mtok
$1.05 / $4.20
makora/fp4fp4
Together84.94%
Down24h 94.67%
7d
96.78%
30d
96.76%
p50
329ms◆ (best)
p99
40.2s
Throughput
112 tps◆ (best)
Context
1.0M
$/Mtok
$1.40 / $4.40
together
Decart2.13%
Down24h 98.19%
7d
99.30%
30d
97.95%
p50
1.22s
p99
21.2s
Throughput
95 tps
Context
1.0M
$/Mtok
$1.19 / $3.74
decart/fp4fp4
Baidu—
No data24h 100%
7d
100%◆ (best)
30d
100%◆ (best)
p50
783ms
p99
1.30s◆ (best)
Throughput
77 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
baidu/fp8fp8
BaseTen—
No data24h 95.88%
7d
99.21%
30d
99.59%
p50
1.23s
p99
34.8s
Throughput
62 tps
Context
1.0M
$/Mtok
$2.10 / $6.60
baseten/fp8fp8
Cloudflare—
No data24h 99.95%
7d
99.78%
30d
99.72%
p50
2.77s
p99
68.0s
Throughput
35 tps
Context
1.3M
$/Mtok
$1.40 / $4.40
cloudflare
Crusoe—
No data24h —
7d
—
30d
99.41%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$1.40 / $4.40
crusoe/fp8fp8
DeepInfra—
No data24h —
7d
—
30d
88.65%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$1.20 / $4.00
deepinfra/bf16bf16
DeepInfra—
No data24h —
7d
—
30d
95.08%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$1.20 / $4.00
deepinfra/fp8fp8
Fireworks—
No data24h —
7d
—
30d
—
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$2.10 / $6.60
fireworks/us
GMICloud—
No data24h 98.53%
7d
99.26%
30d
98.74%
p50
2.62s
p99
14.4s
Throughput
43 tps
Context
1.0M
$/Mtok
$1.05 / $3.30
gmicloud/fp8fp8
Io Net—
No data24h —
7d
100%◆ (best)
30d
99.72%
p50
—
p99
—
Throughput
—
Context
262K
$/Mtok
$1.13 / $4.32
io-net/fp4fp4
Mistral—
No data24h 98.84%
7d
99.34%
30d
99.34%
p50
4.03s
p99
209s
Throughput
46 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
mistral/nvfp4nvfp4
Morph—
No data24h —
7d
100%◆ (best)
30d
95.37%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$1.00 / $3.41
morph/fp4fp4
SiliconFlow—
No data24h 99.60%
7d
99.22%
30d
99.29%
p50
2.00s
p99
14.7s
Throughput
47 tps
Context
1.0M
$/Mtok
$1.12 / $3.52
siliconflow/fp8fp8
StreamLake—
No data24h —
7d
—
30d
100%◆ (best)
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$1.11 / $3.50
streamlake
StreamLake—
No data24h —
7d
99.65%
30d
98.85%
p50
—
p99
—
Throughput
—
Context
1.0M
$/Mtok
$0.700 / $2.20◆ (best)
streamlake/fp8fp8
Venice—
No data24h 97.42%
7d
98.51%
30d
96.21%
p50
695ms
p99
4.40s
Throughput
48 tps
Context
1.0M
$/Mtok
$1.40 / $4.40
venice

Measured 3m 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, averaged over the hour.

Throughput

Median tokens per second.