#06SJun 20, 2026Evidence A

GLM-5

Name: GLM-5 multimodal review
Item: GLM-5
Rating: 92
Author: CN Multimodal Tier List

Zhipu AI · Undisclosed (MoE, trained on Ascend)

The all-rounder with the deepest domestic-chip support

GLM-5 was trained entirely on Ascend; GLM-5V-Turbo ranks near the top of SuperCLUE-VLM. Chinese documents, on-prem government work and agent workflows are its three cards.

TextImageVideoNative

Last updated Oct 18, 2026Editorial scoring

Bottom line

When domestic silicon is mandatory, GLM-5 is often the only option that satisfies chip, compliance and quality at once. Raw capability trails Doubao and Qwen flagships, but on-prem delivery experience and Ascend support are real moats.

Eight-dimension breakdown

editorial mapping · 0–100

GLM-592.0

Visual perception & grounding20%

Document parsing & Chinese OCR18%

Chart & document reasoning14%

Video & long-horizon reasoning14%

GUI agents & tool use14%

Generation & editing6%

Chinese context & culture6%

Ecosystem, cost & availability8%

Weights: perception 20 / documents 18 / charts 14 / video 14 / agents 14 / generation 6 / Chinese 6 / ecosystem 8. Definitions and sources on the methodology page.

Public benchmarks

as reported, not converted

Benchmark / metric	Value	Note
GLM-5.1 SWE-Bench Pro	70%	Official disclosure
GLM-4.5V 视觉基准	42 项	Best open at its scale (official)
SuperCLUE-VLM	前列	GLM-5V-Turbo listed

Sources:

智谱开放平台 SuperCLUE-VLM 榜单

Public scores come from different versions and prompts and are not interchangeable with our composite; they are one evidence input.

Scenario mapping

five stress tests, one scale

Scenario	Score	Observation
Chinese receipt OCR		95 · Especially strong on official docs and tables
Chart data extraction		91 · Stable on standard charts
30-min video QA		86 · Temporal localization is the weak spot
Mobile GUI control		90 · Well-formed tool calls
Chinese cultural context		95 · Strong policy and cultural context

Strengths and limitations

Strengths

Full Ascend stack support and the most on-prem case studies.
Strong Chinese official writing, academic prose and long-form structure.
Stable agent orchestration with disciplined function-call formats.

Limitations

Video temporal reasoning trails the leaders by about half a generation.
Flagship closed; open versions lag behind.
Global ecosystem and multilingual coverage trail Qwen.

Price and real cost

List price

¥10/M tokens

≈ ¥10 / 1M output tokens. For 1,000 images at 1,024px, ~1,100 image tokens each and 300 output tokens, the monthly cost is about ¥6.0 (assuming input at 25% of output price).

Use your volume

About GLM-5

Is GLM-5 open source?

Licence: Commercial + on-prem license. No public weights; access is via vendor API or an on-prem licence.

How much does GLM-5 cost?

≈ ¥10 / 1M output tokens. It ranks #6 with a composite of 92.0. Real bills depend on resolution, tiling, caching and volume discounts.

What is GLM-5 best for?

Government/enterprise on-prem and Xinchuang; Chinese official documents and research reports; Complex agent workflows。

Why does GLM-5 score 92.0?

The composite weights eight dimensions: perception 20%, documents 18%, charts 14%, video 14%, agents 14%, generation 6%, Chinese 6%, ecosystem 8%. Its strongest dimension is Document parsing & Chinese OCR; its weakest is Video & long-horizon reasoning.

Personal ranking · Not an official score

Every ranking, score and comparison on this page is an independent personal ranking, not an official score, and represents no vendor, institution or regulator. Scores are a subjective mapping of public evidence under our own weights — change the weights and the order changes. Disagree, or spotted a factual error? Contact the site owner: jizhi2024@foxmail.com — we check each report and log corrections in the changelog.