GLM-5.3 FlashX is the high-speed variant of Zhipu AI's GLM-5.3 Flash. It keeps the same hybrid sparse-plus-linear attention architecture and the same 1,048,576-token context window, but is tuned for throughput — Zhipu reports up to 200 tokens per second.
It sits in the GLM-5.3 family alongside GLM-5.3 and GLM-5.3-Flash, trading some of the flagship's depth for speed and a lower price.
Throughput first
Up to 200 tokens per second, vendor-reported, on the same architecture as GLM-5.3 Flash. Built for paths where latency is the constraint.
A million tokens of context
1,048,576 tokens in, up to 131,072 back — enough for whole repositories or long transcripts without chunking.
Agentic plumbing
Parallel tool calling, structured (JSON schema) output and streaming, which is what agent loops need to plan, call tools and iterate.
Cheap cached input
At $0.0975 per million, cached input is roughly a fifth of the standard input rate — material for workflows that re-send the same long prefix.
Reasoning tokens are billed at the output rate. Structured tool calls carry an additional charge of $0.013 per call.
Latency-bound chat
Interactive assistants where time-to-answer matters more than the last increment of quality.
High-volume pipelines
Classification, extraction and summarisation run continuously rather than occasionally.
Long-context work
The 1M window removes most of the need for retrieval scaffolding.
Agents with tools
Parallel tool calling and structured output for multi-step workflows.
Source: https://api.aimlapi.com/v1/models
Measured on the AI/ML API gateway across live traffic. Figures are medians unless stated. Last updated 2026-10-12.
import requests r = requests.post( "https://api.aimlapi.com/v1/chat/completions", headers={"Authorization": "Bearer " + AIMLAPI_KEY}, json={ "model": "z-ai/glm-5.3-flashx", "messages": [ { "role": "user", "content": "Hello!" } ] }, ) print(r.json())
const r = await fetch("https://api.aimlapi.com/v1/chat/completions", { method: "POST", headers: { Authorization: `Bearer ${process.env.AIMLAPI_KEY}`, "Content-Type": "application/json", }, body: JSON.stringify({ "model": "z-ai/glm-5.3-flashx", "messages": [ { "role": "user", "content": "Hello!" } ] }), }); console.log(await r.json());
curl -X POST https://api.aimlapi.com/v1/chat/completions \ -H "Authorization: Bearer $AIMLAPI_KEY" \ -H "Content-Type: application/json" \ -d '{"model":"z-ai/glm-5.3-flashx","messages":[{"role":"user","content":"Hello!"}]}'
OpenAI-compatible — swap the base URL and it works with your existing SDK.
| Type | Price |
|---|---|
| Input | |
| Output | |
| Cached input |
Reasoning tokens are billed at the output rate. Structured tool calls carry an additional charge of $0.013 per call.
GLM 5.3 FlashX has a 1,048,576 token context window and can return up to 131,072 tokens.
FlashX is the high-speed variant. It runs on the same hybrid sparse-plus-linear attention architecture and the same 1M-token context window as GLM-5.3 Flash, but is tuned for throughput — Zhipu reports up to 200 tokens per second.
GLM 5.3 FlashX is priced at input $0.481 / 1M tokens, output $1.625 / 1M tokens, cached input $0.0975 / 1M tokens. Reasoning tokens are billed at the output rate, and structured tool calls carry an extra $0.013 per call.
It accepts images and text as input and returns text. File input is supported as well.
Yes. It supports function calling including parallel tool calls, structured (JSON schema) output, streaming and reasoning. Note that structured tool calls are billed separately at $0.013 per call.
Zhipu AI reports up to 200 tokens per second. That figure is vendor-reported and has not been independently reproduced.
GLM 5.3 FlashX was built by Zhipu AI as part of the GLM-5.3 family, alongside GLM-5.3 and GLM-5.3 Flash. It became available on September 18, 2026.
It is served on the OpenAI-compatible chat completions endpoint. Point any OpenAI SDK at https://api.aimlapi.com/v1 and set the model to z-ai/glm-5.3-flashx. The aliases glm-5.3-flashx and zhipu/glm-5.3-flashx also resolve to it.
