GLM 5.3
z-ai/glm-5.3- Z.ai
- Family · GLM 5
- Released 18 Aug 2026
- Text → Text
- Reasoning · adjustable
- Tools · JSON
Spec & list price
Context1M
Max output944K
Input / 1M$0.08
Output / 1M$3.6045× input
Scores · standing
Intelligence44.8#21/489
Coding74.8#14/196
Usage
Tokens / day403.1B▼ 19% 7d
Usage rank#13
Requests / day4.2M
Tool calls / req0.55
Per request · 30 days
Avg input65.7K4.4× median
Avg output1.0K1.5× median
Cache hit91%1.3× median
Reasoning share72%1.3× median
Tool-call failures0.7%0.4× median
Consumption
Peak 981.0B 15 OctLow 215.6B 27 SepRank today #13
News & trends
114 mentions this week ▲ 500%this model · videos, posts, articles · 30 days
Trending · 14 days
- GLM 5.3 can autonomously compromise recent versions of Mozilla Firefox626 upvotes
- @EpochAIResearch on X86.0K views
- Hermes and REA and GLM 5.3 = Reverse engineering TikTok app for less than $2019 points
- AI OFM FULL COURSE #5 — PHONE FARM AUTOMATION With GLM 5.369.7K views
- Cross-request KV cache exposure in the GLM-5.3 Workers AI model (155 responses)178 points
- Mixing Turkish, English, Russian and Chinese all together, thank you GLM 5.3 Flash Free 🤣🤣149 upvotes
- Project Maya Deepseek V4.1 Flash 1017 + GLM 5.3 Local Ai Review58.2K views
News
- Hermes and REA and GLM 5.3 = Reverse engineering TikTok app for less than $20scraping.club · 29 Oct
- GLM-5.3 Sometimes Considers How Train-Time Actions Shape Future Behaviorlesswrong.com · 27 Oct
- Cross-request KV cache exposure in the GLM-5.3 Workers AI model (155 responses)blog.cloudflare.com · 26 Oct
- Together Link: open models in the harness you already use. Start with one command today.Together AI · 5 Oct
Benchmarks
bar = share of the top score · Benchmark data: Artificial AnalysisSpeed · median across 47 providers, latest day
Output speed#115/33484 tok/s1.3× faster than median
Fastest provider168 tok/sbest of 47 providers
First token#1160.76 s1.4× quicker than median
First answer token#7526 sthinking included
#6 cheapest of 14 with a similar score
Value
Intelligence Index
Blended price per 1M tokens (log)
Family progress
Intelligence Index
- Family releases44.8
Technical details
- Reasoning
- on · max
- Tokenizer
- Other
- License
- other
- Moderation
- none
- Downloads · likes
- 2.5M · 2.4K
Reasoning effort
- max · default
- high
- low
Request parameters 21
- tools
- tool_choice
- parallel_tool_calls
- response_format
- structured_outputs
- reasoning
- reasoning_effort
- include_reasoning
- temperature
- top_p
+11 more
- top_k
- min_p
- max_tokens
- stop
- seed
- frequency_penalty
- logit_bias
- logprobs
- presence_penalty
- repetition_penalty
- top_logprobs
Interesting facts
- 2.4× the output room of similar models (943.7K against a 393.2K median).
- The most used of 4 GLM 5 models by tokens last month.
- Released 76 days ago, the newest of 4 GLM 5 models.
When to use
Best for
- Hard coding and reasoning tasks
- Banking-style customer tool workflows
- Very long documents and codebases
- Large outputs such as full reports
Think twice for
- Browser-driving agents
- Image input
- Cheapest high-volume chat
- Self-hosting (weights not confirmed open)
Research papers
- ATLAS-Bench: An Agentic Tool-Use Benchmark for Local Area Search29 Oct
- Agent-Conditioned Routing: Diverse Reasoning Paths Within One Frozen Model27 Oct
- Correctness Conditioning as Selection in Chain-of-Thought Faithfulness Evaluation27 Oct
- CyberDual: AI Attackers versus AI Code Reviewers in Real Software Maintenance25 Oct
- DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems20 Oct
- CoEvo-RSI: When Harnesses Evolve, Which Models Should Follow?18 Oct
- Can LLMs predict the outcomes of machine learning experiments?15 Oct
- MLExpert: Towards a Generalist Agentic Model for Autonomous Machine Learning Engineering15 Oct
Social
Most watched and discussed · 90 days
- AI OFM FULL COURSE #5 — PHONE FARM AUTOMATION With GLM 5.3Manila AI OFM · 69.7K views · 28 Oct
- Project Maya Deepseek V4.1 Flash 1017 + GLM 5.3 Local Ai ReviewDigital Spaceport · 58.2K views · 20 Oct
- DeepSeek 4.1 Flash e GLM 5.3 AGORA É GRÁTIS. Qual a Pegadinha?Robson Oliveira - Vibe Coding · 25.8K views · 27 Oct
- 763B AI Is Finally Running at Home — Project Maya Runs DeepSeek V4.1 & GLM 5.3 on 24GBRun Local AI · 15.4K views · 21 Oct
- 500M FREE AI Tokens! GLM-5.3 + GLM-5.3-Flash in ZCode (Limited Time)Code With Yousaf · 6.6K views · 24 Oct
- 2B FREE AI Tokens! 🔥 GLM 5.3 & GLM 5.3 Flash + FREE API Key 4 Platforms (Tested)CodeWithUsman · 3.2K views · 25 Oct
- This coding agent runs Kimi K3.1, DeepSeek V4, GLM 5.3 & Qwen (Cline)ByteGrad · 2.5K views · 31 Oct
- Get a Free AI API Key Daily | GLM-5.3 Flash & DeepSeek V4.1 FlashAkash Singh · 2.1K views · 31 Oct
- GLM 5.3: El modelo al que Anthropic le tiene MIEDO. 😳Charly Prado · 1.6K views · 31 Oct
- DeepSeek V4.1 1017 vs GPT 6.1 Terra vs GLM 5.3 FlashAI GAME LAB · 1.3K views · 31 Oct
- $4 400 — и у нейросети нет защиты: Anthropic проверила китайскую GLM-5.3 🔓 #shortsIvanArtAstra · 1.3K views · 24 Oct
- 10 Million FREE Tokens for AI Coding 🤯 | GLM-5.3, DeepSeek V4.1, Qwen 3.8 Max & MiniMax M3Oxg Tech · 1.3K views · 28 Oct
Same family
GLM 5- 33.7
- 26.1
- 27.9
Alternatives
other vendors · similar capability- 42.7
- 43.6
- 43.4
- I41.1
Using GLM 5.3 at volume?
Sign in to ask for a volume discount. You will continue in the OSMEXO dashboard.