GLM-5.3-Flash API pricing and cost calculator

How this is calculated

A request with 100,000 input tokens and 5,000 output tokens costs $0.015 for uncached input and $0.0025 for output, totaling $0.0175. Reading half the input from cache reduces input to $0.009 and the request total to $0.0115. At 100 monthly requests, the starting chart is $1.15. At 100,000 identical requests, that is $1,150 with half the input cached or $1,750 without cache hits. A fully cached 100,000-token prompt plus 5,000 generated tokens costs $0.0055. The 80% input discount doesn't apply to output. Output costs become more visible when the model generates long explanations or reasoning traces, so use the provider's billed output usage rather than the length of the final answer alone. Z.ai's model guide lists a 1M-token context window and maximum output of 128K tokens. Those are capacity limits, not a recommended request size. Sending more context raises token volume even when the rate remains the same. These are normal-speed real-time API prices. A coding plan's off-peak points policy describes subscription usage and needs its own calculation. Z.ai lists web search at $0.01 per use, so 100,000 such uses add $1,000 beyond the token estimate.

Verdict

GLM-5.3-Flash is a candidate for coding, document processing, and tasks that use visual input. Test its results on the work you plan to automate and measure complete request usage. Keep reusable instructions near the start of the prompt, check actual cache hits, and give the model a suitable output budget. Compare another provider using the same workload and delivery timing. Cache-heavy requests can rank differently from requests with little reused input, so one uncached example shouldn't decide every deployment.