Home / Models / Z.ai / GLM 5.3

GLM 5.3 pricing and routing profile

GLM 5.3 from Z.ai lists at $0.325 per million input tokens and $4.40 per million output tokens, with a 1,310,720-token context window. Prices from OpenRouter's catalogue on 2026-09-28.

Input, per 1M tokens$0.325
Output, per 1M tokens$4.40
Cached input, per 1M$0.26
Context window1.31M
Confidence signalAvailable

Tool callingStructured outputReasoning controlLog probabilities

Cost per 1,000 requests

Request shapeList cost
Classification or routing (600 tokens in, 10 out)$0.239
Chat reply (2,000 in, 400 out)$2.41
Retrieval-augmented answer (8,000 in, 500 out)$4.80

Before prompt caching. Try your own shape in the cost calculator.

Can Gimbix route to it first?

Yes. 22 of 40 tracked hosts return token log probabilities for GLM 5.3, which is the confidence signal Gimbix uses to let a cheaper model answer first.

Who serves GLM 5.3

HostInput / output per 1MLog probabilitiesUptime, last 30 min
Wafer$0.325 / $4.40yes99.9%
Relace$0.33 / $2.50no99.6%
Inceptron$0.442 / $3.28yes99.8%
Morph$0.452 / $2.81yes98.2%
Io Net$0.50 / $1.90yes100.0%
DeepInfra$0.563 / $2.50no99.6%
Reka$0.683 / $2.57yes100.0%
InferenceNet$0.684 / $2.28no98.1%
Sail Research$0.77 / $4.00yes100.0%
Sail Research$0.77 / $4.00yes100.0%
Novita$0.783 / $2.46no100.0%
DigitalOcean$0.91 / $2.86yes98.3%
Phala$0.98 / $3.08no100.0%
GMICloud$0.98 / $3.08no—
Makora$1.05 / $4.20yes98.3%
SiliconFlow$1.12 / $3.52no—
Alibaba$1.19 / $3.74yes100.0%
Decart$1.19 / $3.74no100.0%
Friendli$1.26 / $3.96no100.0%
AkashML$1.30 / $4.40yes—
Mistral$1.40 / $4.40yes99.6%
Wafer$1.40 / $4.40yes99.8%
Baidu$1.40 / $4.40no100.0%
BaseTen$1.40 / $4.40no90.8%
Mistral$1.40 / $4.40yes99.8%
Crusoe$1.40 / $4.40no99.2%
PrimeIntellect$1.40 / $4.40yes100.0%
Venice$1.40 / $4.40no95.1%
Together$1.40 / $4.40yes99.6%
Parasail$1.40 / $4.40yes99.7%
Modal$1.40 / $4.40yes99.9%
BaseTen$1.40 / $4.40no—
Fireworks$1.40 / $4.40yes99.7%
Cloudflare$1.40 / $4.40yes—
AtlasCloud$1.40 / $4.40no—
Z.AI$1.40 / $4.40no100.0%
Mistral$1.54 / $4.84yes—
BaseTen$2.10 / $6.60no100.0%
BaseTen$2.10 / $6.60no—
Alibaba$2.80 / $8.80yes—

A host that lists log probabilities may still omit them; Gimbix checks each one before relying on it.

Cheaper models that can answer first

Models priced below GLM 5.3 that expose a confidence signal. Whether they answer your traffic as well is what a Gimbix pilot measures.

ModelInput / output per 1MCheaper on a chat reply
DeepSeek V4 Pro 0813$0.227 / $4.2011%
Qwen3.6 27B$0.32 / $3.2020%
Qwen3 235B A22B Thinking 2507$0.23 / $2.3043%
Qwen3 VL 30B A3B Thinking$0.20 / $2.4044%
Qwen3.5-122B-A10B$0.26 / $2.0844%
Qwen3.5 Plus 2026-04-20$0.30 / $1.8045%

Questions

How much does GLM 5.3 cost?

GLM 5.3 lists at $0.325 per million input tokens and $4.40 per million output tokens on OpenRouter as of 2026-09-28. A typical chat reply of 2,000 input and 400 output tokens costs about $2.41 per 1,000 requests.

What is the context window of GLM 5.3?

1,310,720 tokens.

Can a cheaper model replace GLM 5.3?

Sometimes. Gimbix proves it on your own traffic before routing: the cheaper model must match GLM 5.3 on one half of your requests and make at most three answers worse on the other.