Is Z.ai GLM 5.3 available with Zero Data Retention on Elosia?

Z.ai GLM 5.3 runs on Elosia's standard backend; ZDR is available on many other models in the catalog.

How much does Z.ai GLM 5.3 cost on Elosia?

Z.ai GLM 5.3 is billed by usage at $0.079 per 1M input tokens and $1.4 per 1M output tokens, within your Elosia plan.

What is Z.ai GLM 5.3's context window?

Z.ai GLM 5.3 supports a context window of up to 1,048,576 tokens.

Is Z.ai GLM 5.3 hosted in the EU?

Yes, Z.ai GLM 5.3 is available on Elosia's EU-hosted (EdenAI) backend for enterprise organizations.

When should you choose Z.ai GLM 5.3?

Z.ai GLM 5.3 is a flagship model, best for the most demanding reasoning, agentic and long-context tasks.

Z.ai GLM 5.3

Z.aiFlagship

ThinkingTool UseStructured Output

About this model

GLM-5.3 is a large-scale reasoning model from Z.ai, built for complex software engineering and long-horizon agent tasks. It supports text input and output with a 1M-token context window, and improves...

Performance Tier

Flagship

Z.ai GLM 5.3 is a flagship model from Z.ai : the most capable in their lineup.

Best-in-class model from this provider. Highest performance across benchmarks, ideal for demanding tasks.

Pricing

This model is included in Elosia plans

Affordable

Low cost. Suitable for sustained use and high-volume interactions.

Type	per 1M tokens
Input (prompt)	$0.079
Output (completion)	$1.40
Cache read	$0.078

Capabilities

Context Length1.0M

Max Output Tokens944K

TokenizerOther

Inputtext

Outputtext

Release DateAugust 18, 2026

Benchmarks

General Intelligence

MMLU

Not reported

GPQA Diamond

91.7%

Mathematics

MATH-500

Not reported

Programming

HumanEval

Not reported

SWE-bench Verified

Not reported

Reasoning

IFEval

Not reported

Humanity's Last Exam

42.3%

Agentic

Terminal-Bench 2.1

88.2%

Where does Z.ai GLM 5.3 stand?

Compare its performance index against every other model.

View the performance leaderboard

Recommended Use Cases

CodingAnalysisResearch

Strengths

Mixture of experts of about 750B parameters with 40B active, whose gains come entirely from scaled post-training for complex software engineering and long-horizon agents
1M-token context window with up to 128K of output, and three reasoning effort levels
Cyber capability evaluated on both vulnerability discovery and exploitation, on real open-source projects
Open weights in BF16 and FP8, with official serving recipes for vLLM, SGLang and Ascend NPUs
Tool calling with parallel calls and structured JSON output

Limitations

Text only: no image or video input
Reasoning is always on and defaults to maximum effort, which adds latency and tokens on simple calls
Custom license rather than MIT: operators of a model-as-a-service business above a revenue threshold must pass a Z.ai security review before commercial use
Headline results are self-reported; Z.ai publishes no GPQA Diamond or no-tools Humanity's Last Exam score, and the ones shown come from third-party tests