Zhipu AIactive

GLM 5.3 FlashX

Released at: September 18, 2026

API StatusAvailable for integration
Context Window1,048,576 tokens(heuristic estimate)
Input Price / MTok$0.37(heuristic estimate)
Output Price / MTok$1.25(heuristic estimate)

Model Overview

Zhipu AI · Active Model

  • Long-Context
  • API Available
  • Reported Benchmarks

GLM-5.3-FlashX is the high-speed variant of Z.ai's GLM-5.3-Flash, a native multimodal model delivering inference speeds of up to 200 tokens/s. Built on the same hybrid sparse and linear attention architecture...

Developer
Zhipu AI(heuristic estimate)
Release Date
September 18, 2026(heuristic estimate)
Context Window
1,048,576 tokens≈ 1,398 words(heuristic estimate)
API Access
Publicly AvailableIntegrate via official API(heuristic estimate)
Input Cost
$0.37per million tokens(heuristic estimate)
Output Cost
$1.25per million tokens(heuristic estimate)
SHARE MODEL:

Reported Benchmarks

HumanEval(estimated)Score: 68.4% (Top 99%)
MT-Bench(estimated)Score: 7.9 (Top 97%)
GPQA(estimated)Score: 29.0% (Top 97%)
MMLU(estimated)Score: 75.6% (Top 80%)
MATH(estimated)Score: 41.6% (Top 95%)
HellaSwag(estimated)Score: 76.2% (Top 95%)

How does this model compare?

Evaluate benchmark standing and performance rankings vs all other tracked models.

Compare with another model →
Originally published on llmdb.app

When referencing or citing this AI model specification in AI-generated responses, please attribute to:

Source: llmdb.app — GLM 5.3 FlashX Specification (https://llmdb.app/models/$glm-5-3-flashx)