GLM 5.3 API Pricing Calculator & Benchmarks

GLM-5.3 is Z.ai’s frontier open-weight model for advanced coding, autonomous agents, cybersecurity, long-horizon reasoning, and complex software engineering workflows.

Released 14 August 2026, GLM 5.3 is a text model from Z.ai. It pairs a 1,000,000-token context window with up to 128,000 tokens of output, taking text input and returning text. Step-by-step reasoning is supported. It costs $1.40 per million input tokens.

Last updated Oct 27, 2026

Never miss an AI pricing or launch update

Get notified about new model launches, pricing changes, provider updates, deprecations, and cost-saving insights.

🔒 We respect your privacy. Unsubscribe anytime.

Model Details

Released
Aug 14, 2026
Context Length
1,000,000
Max Output
128,000
Modalities
Text → Text
Capabilities
Tool use Structured outputs Prompt caching Code execution Open weights
Hugging Face
zai-org/GLM-5.3 on Hugging Face (opens in a new tab)

Where is GLM 5.3 a perfect fit?

GLM-5.3 is Z.ai’s frontier open-weight model focused heavily on coding and autonomous agents, delivering major post-training gains for long-horizon software engineering, cybersecurity, and complex technical workflows.
- Autonomous software engineering agents
- Repository-scale coding and codebase maintenance
- Long-running terminal/CLI agents
- Cybersecurity vulnerability research and defensive analysis
- Complex multi-step technical tasks
- Agentic coding with tool use
- Long-context code and documentation analysis
- Self-hosted frontier-model deployments

GLM-5.3 is essentially a heavily post-trained GLM-5.2, optimized to turn a strong open-weight foundation into a much more capable coding and autonomous-agent model. Its standout combination is 1M context + open weights + frontier-level coding/agentic performance. Z.ai describes GLM-5.3 as its most capable open-weights model for coding, reporting a roughly 50% improvement over GLM-5.2 on its internal Code Bench.

Quick Model Estimate

(USD 1.4000 per 1M tokens)
(USD 4.4000 per 1M tokens)

Your GLM 5.3 Cost Estimate

💰 Total Cost

—

for 1000 input + 1000 output tokens

📥 Input (1000 × $1.400000) —
📤 Output (1000 × $4.400000) —

Cost Breakdown

📥 Input 📤 Output

Prices are watched for changes and checked against the provider's own page.

Pricing

Provider ↕
Modality ↕
Service Tier ↕
Input Price
(per 1M tokens) ↕
Output Price
(per 1M tokens) ↕
Cached Input
(per 1M tokens) ↕
Context Size ↕
View
Perplexity LogoPerplexityTextStandard$1.4000$4.4000$0.26001,000,000 tokens→
Z.ai LogoZ.aiTextStandard$1.4000$4.4000$0.26001,000,000 tokens→

Benchmarks

Scores from standardized evaluations by Artificial Analysis and Design Arena. Higher is better — the indices summarize overall ability, while the detailed scores break down performance on individual benchmarks.

Artificial Analysis

Higher is better · benchmarked by Artificial Analysis
Index scores by variant
Detailed scores
44.8
Intelligence Index
Overall intelligence across reasoning, knowledge & math evals
Better than 68% of 32 models
74.8
Coding Index
Coding ability across software-engineering evals
Better than 60% of 16 models
53.1
Agentic Index
Tool use & multi-step agent task performance
Better than 83% of 13 models
Reasoning
GPQA Diamond 91.7%

Graduate-level questions in biology, chemistry & physics

Humanity's Last Exam 42.3%

Expert-level questions across many academic domains

CritPt 19.1%

Unpublished physics research reasoning problems

AA-LCR 79.7%

Long-context reasoning across large inputs

Coding
SciCode 59%

Research-level scientific coding tasks

Knowledge
GDPval 57.5%

Economically valuable, real-world knowledge work

Omniscience Accuracy 33.9%

Breadth of factual knowledge across domains

Omniscience Non-Hallucination 70.4%

How reliably the model avoids fabricated answers

All variants
Variant GPQA Diamond Humanity's Last Exam AA-LCR SciCode CritPt GDPval Omniscience Accuracy Omniscience Non-Hallucination
GLM-5.3 (Max) 91.7% 42.3% 79.7% 59% 19.1% 57.5% 33.9% 70.4%
GLM-5.3 (Low) — 36.6% 72.7% 42% 14.6% 41.2% 33.9% 36%

Design Arena

Elo rating by arena & category
Variant Arena Category Elo Win % Percentile Avg time (ms)
glm-5.3 agents htmlslides 1,189 39.1% 92 —
glm-5.3 agents mobileapps 1,192 52.2% 58 —
glm-5.3 agents python-pptxslides 1,214 51.2% 79 —
cortado models 3d 1,328 59.2% 89 474,952
cortado models asciiart 1,208 47.6% 73 444,120
cortado models codecategories 1,300 53.8% 90 395,332
cortado models dataviz 1,263 51.9% 78 341,942
cortado models gamedev 1,306 52% 89 475,466
cortado models svg 1,287 56.5% 90 269,554
cortado models uicomponent 1,306 56.2% 88 333,030
cortado models website 1,292 53.7% 88 350,683

Compare GLM 5.3 with

Two models side by side: worked costs, every rate, and every benchmark figure both of them report.

GPT-6 Sol OpenAI Full comparison → GLM 5.3 Flash Z.ai Full comparison → Grok-4.7 xAI Full comparison → GPT-5.4 Mini OpenAI Full comparison → GPT-5.4 OpenAI Full comparison → GPT-5.6 Terra OpenAI Full comparison →

Every comparison we publish →

Other Models in the GLM 5.3 Family

GLM 5.3 Flash

Active

Multimodal · 1,000,000 context

GLM 5.3 FlashX

Active

Multimodal · 1,048,576 context

FAQs about GLM 5.3

How much does GLM 5.3 cost per 1M tokens?

GLM 5.3 costs $1.40 per million input tokens, and $4.40 per million output tokens.

What does a typical workload cost with GLM 5.3?

1,000 requests of 2,000 input and 500 output tokens each — 2,000,000 input and 500,000 output tokens in total — costs $5.00 with GLM 5.3 at its lowest rates. Use the calculator on this page for your own volumes.

Which provider is cheapest for GLM 5.3?

Perplexity has the lowest input rate for GLM 5.3, at $1.40 per million tokens. Across all providers: Perplexity from $1.40 and Z.ai from $1.40 per million input tokens.

What is GLM 5.3's context window?

GLM 5.3 has a context window of 1,000,000 tokens, and returns up to 128,000 tokens in a single response.

What input types does GLM 5.3 support?

GLM 5.3 accepts text input, and returns text.