GLM-5.3-Flash API pricing in 2026: what 1M tokens costs

GLM-5.3-Flash on Z.ai's list price (official price page, checked 24 October 2026): 1M tokens you send cost $0.15, and 1M tokens it writes back cost $0.50. On CheapTokens, MiniMax-M2.7 has live prices, shown in the tables below, and it is cheaper than OpenRouter's reference price in total, with input and output counted together. Below: what that means for a bill you can actually picture.

The prices, side by side#

Z.ai's pricing page (24 October 2026) lists three official prices for GLM-5.3-Flash per 1M tokens: $0.15 to send, $0.03 to send cached tokens and $0.50 to get tokens back. The page shows no publication date, so treat the numbers as "as of 24 October 2026".

OpenRouter's reference price for GLM-5.3-Flash output (23 October 2026) is $0.1985 per 1M tokens. On CheapTokens, MiniMax-M2.7 output is $0.15 per 1M tokens, and input is $0.05. Those two numbers are live: when the price list changes, the page updates itself.

One price per model, at any hour. There are no peak hours and no time-of-day surcharge.

Example 1: a support bot with 1,000 requests#

Say you run a small support bot. Each request carries about 500 tokens in (the question plus a short instruction) and returns about 300 tokens. A thousand requests is 0.5M tokens in and 0.3M out.

On Z.ai's list price that is 0.5 × $0.15 + 0.3 × $0.50 = $0.225. On CheapTokens the same load costs $0.07, and the table below recalculates by itself:

The answer is shorter than the question, but a token of answer costs more: of the $0.225, $0.15 is output. So the output price decides this bill, and MiniMax-M2.7 on CheapTokens has cheaper output than OpenRouter.

Example 2: a coding agent over 200 steps#

Agents behave differently. Every step re-sends the files, the tool results and the history, so each call carries a lot of input and returns only a short action. Take 200 steps with about 20,000 tokens in and 500 out per step.

For this shape of load a good fit is MiniMax-M2.7 (MiniMaxAI/MiniMax-M2.7): it is cheaper than OpenRouter in total, with input and output counted together, and has a 180K context, plenty for a single step. 200 steps come to 4M tokens in and 0.1M out, and on MiniMax-M2.7 that costs $0.215. For reference, the same load at Z.ai's GLM-5.3-Flash list price (24 October 2026) is 4 × $0.15 + 0.1 × $0.50 = $0.65. For the longer story on why history inflates the bill, see why a long AI chat costs more.

What you get with GLM-5.3-Flash on CheapTokens#

• Model id: zai-org/GLM-5.3-Flash
• Context window: 390K tokens
• Up to 8,192 output tokens per answer
• API: OpenAI Chat Completions and Anthropic Messages, at https://api.cheaptokens.dev/v1
• One balance for all models, no peak hours

The current prices of every model are on the pricing page; the GLM family is described on the GLM page.

Send the first request#

Create a key in your account, top up the balance and run this. The key is a placeholder:

curl https://api.cheaptokens.dev/v1/chat/completions \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zai-org/GLM-5.3-Flash",
    "messages": [{"role": "user", "content": "Summarise in one sentence: tokens are billed for input and output separately."}],
    "max_tokens": 200
  }'

The response has a usage block with prompt_tokens and completion_tokens. Multiply them by the prices above and you have the exact cost of that call. Details are in the docs and in your first API request.

Which one to pick#

Want a broader comparison? GLM vs DeepSeek vs MiniMax walks through the three families, and DeepSeek API pricing does the same price breakdown for DeepSeek.