GLM 5.3 Flash

Z.ai2026-08-26● 开源 · MIT🤗 zai-org/GLM-5.3-Flash652.6万2,912

GLM-5.3-Flash 是智谱（Z.ai）推出的原生多模态模型，适用于高效编码与长周期 Agent 任务，其混合稀疏与线性注意力架构在保持长上下文准确性方面表现出色。

多服务商价格对比数据来自 OpenRouter

35 个服务商1 USD ≈ 6.6893 CNY · 更新于 2026-10-22

服务商	量化精度	上下文	输入/M	输出/M	缓存读取/M	近1天可用率
Relace	—	1.05M	¥0.12	¥0.81	¥0.12	99.9%
O OpenInference	fp4	1.05M	¥0.13	¥0.81	¥0.13	99.9%
StreamLake限时 -71%	fp8	1.02M	¥0.29	¥0.97	¥0.06	97.2%
W Wafer	nvfp4	1.05M	¥0.50	¥1.34	¥0.40	100.0%
G GMICloud限时 -50%	fp8	1.05M	¥0.50	¥1.67	¥0.10	98.6%
S Sail Research	fp4	1.05M	¥0.52	¥1.67	¥0.17	100.0%
S Sail Research	fp4	1.05M	¥0.52	¥1.67	¥0.17	100.0%
N Near AI	fp8	1.05M	¥0.54	¥3.34	¥0.08	92.7%
R Reka	—	1.05M	¥0.54	¥10.70	¥0.27	96.7%
Novita限时 -44%	fp8	1.05M	¥0.56	¥1.87	¥0.11	99.9%
I InferenceNet	fp4	1.05M	¥0.80	¥2.60	¥0.21	99.7%
I InferenceNet	fp4	1.05M	¥0.80	¥2.60	¥0.21	98.5%
P Phala限时 -15%	nvfp4	1.05M	¥0.85	¥2.84	¥0.17	99.4%
DigitalOcean	—	1.05M	¥0.88	¥2.93	¥0.20	99.7%
M Makora	—	1.05M	¥0.94	¥3.14	¥0.16	76.6%
P PrimeIntellect	—	1.05M	¥1.00	¥3.34	¥0.20	97.3%
Modal	nvfp4	1.05M	¥1.00	¥3.34	¥0.20	-
BaseTen	fp8	1.05M	¥1.00	¥3.34	¥0.20	-
Crusoe	fp4	1.05M	¥1.00	¥3.34	¥0.20	83.3%
C CoreWeave	nvfp4	1.05M	¥1.00	¥3.34	¥0.13	99.9%
AtlasCloud	fp8	1.05M	¥1.00	¥3.34	¥0.20	95.5%
Fireworks	—	1.05M	¥1.00	¥3.34	¥0.20	99.3%
Friendli	—	1.05M	¥1.00	¥3.34	¥0.20	100.0%
S SiliconFlow	fp8	1.05M	¥1.00	¥3.34	¥0.20	99.0%
Together	—	1.05M	¥1.00	¥3.34	¥0.20	98.1%
Parasail	fp4	1.05M	¥1.00	¥3.34	¥0.20	99.9%
BaseTen	fp8	1.05M	¥1.00	¥3.34	¥0.20	-
Venice	—	1.05M	¥1.00	¥3.34	¥0.20	100.0%
DeepInfra	fp4	1.05M	¥1.00	¥3.34	¥0.20	99.8%
Z.AI	fp8	1.05M	¥1.00	¥3.34	¥0.20	100.0%
I Inceptron	fp8	1.05M	¥1.00	¥4.62	¥0.50	99.9%
N NextBit	fp4	1.05M	¥1.00	¥5.02	¥0.33	84.0%
Morph	fp8	1.05M	¥1.06	¥4.35	¥0.30	98.2%
Parasail	fp4	1.05M	¥1.25	¥4.18	¥0.25	99.7%
Fireworks (US)	—	1.05M	¥1.51	¥5.02	¥0.30	100.0%

Relace

¥0.12

输入/M

¥0.81

输出/M

¥0.12

缓存/M

99.9%

可用率

O

OpenInferencefp4

¥0.13

输入/M

¥0.81

输出/M

¥0.13

缓存/M

99.9%

可用率

StreamLakefp8限时 -71%

¥0.29

输入/M

¥0.97

输出/M

¥0.06

缓存/M

97.2%

可用率

W

Wafernvfp4

¥0.50

输入/M

¥1.34

输出/M

¥0.40

缓存/M

100.0%

可用率

G

GMICloudfp8限时 -50%

¥0.50

输入/M

¥1.67

输出/M

¥0.10

缓存/M

98.6%

可用率

S

Sail Researchfp4

¥0.52

输入/M

¥1.67

输出/M

¥0.17

缓存/M

100.0%

可用率

S

Sail Researchfp4

¥0.52

输入/M

¥1.67

输出/M

¥0.17

缓存/M

100.0%

可用率

N

Near AIfp8

¥0.54

输入/M

¥3.34

输出/M

¥0.08

缓存/M

92.7%

可用率

R

Reka

¥0.54

输入/M

¥10.70

输出/M

¥0.27

缓存/M

96.7%

可用率

Novitafp8限时 -44%

¥0.56

输入/M

¥1.87

输出/M

¥0.11

缓存/M

99.9%

可用率

I

InferenceNetfp4

¥0.80

输入/M

¥2.60

输出/M

¥0.21

缓存/M

99.7%

可用率

I

InferenceNetfp4

¥0.80

输入/M

¥2.60

输出/M

¥0.21

缓存/M

98.5%

可用率

P

Phalanvfp4限时 -15%

¥0.85

输入/M

¥2.84

输出/M

¥0.17

缓存/M

99.4%

可用率

DigitalOcean

¥0.88

输入/M

¥2.93

输出/M

¥0.20

缓存/M

99.7%

可用率

M

Makora

¥0.94

输入/M

¥3.14

输出/M

¥0.16

缓存/M

76.6%

可用率

P

PrimeIntellect

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

97.3%

可用率

Modalnvfp4

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

-

可用率

BaseTenfp8

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

-

可用率

Crusoefp4

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

83.3%

可用率

C

CoreWeavenvfp4

¥1.00

输入/M

¥3.34

输出/M

¥0.13

缓存/M

99.9%

可用率

AtlasCloudfp8

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

95.5%

可用率

Fireworks

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

99.3%

可用率

Friendli

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

100.0%

可用率

S

SiliconFlowfp8

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

99.0%

可用率

Together

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

98.1%

可用率

Parasailfp4

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

99.9%

可用率

BaseTenfp8

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

-

可用率

Venice

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

100.0%

可用率

DeepInfrafp4

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

99.8%

可用率

Z.AIfp8

¥1.00

输入/M

¥3.34

输出/M

¥0.20

缓存/M

100.0%

可用率

I

Inceptronfp8

¥1.00

输入/M

¥4.62

输出/M

¥0.50

缓存/M

99.9%

可用率

N

NextBitfp4

¥1.00

输入/M

¥5.02

输出/M

¥0.33

缓存/M

84.0%

可用率

Morphfp8

¥1.06

输入/M

¥4.35

输出/M

¥0.30

缓存/M

98.2%

可用率

Parasailfp4

¥1.25

输入/M

¥4.18

输出/M

¥0.25

缓存/M

99.7%

可用率

Fireworks (US)

¥1.51

输入/M

¥5.02

输出/M

¥0.30

缓存/M

100.0%

可用率