GLM-5.3
Post-training-only upgrade that turned GLM into a serious open terminal agent.
Overall Intelligence
#32
of 56
13.1 pts behind #1
0
races led
0
weeks at #1
75.6
score
- Released
- Aug 14, 2026
- 2 mo ago
- Context
- 1M
- 128K max output
- Price
- $1.4 / $4.4
- per 1M in/out · $2.15 blended
- Speed
- 76 tok/s
- 2.5s to first token
- Modalities
- text
- outputs text
- Weights
- Open
- Open weights (GLM-5.2 base)
The 30-second version
A post-training-only upgrade that turned GLM into a serious open terminal agent.
- 1Strong open-weights terminal and DevOps agent.
- 2$1.40 / $4.40 via API, or self-host.
- 3Proof that post-training alone can move agentic skills a lot.
- 4Middling on writing and multimodal tasks.
What you need to know
Purpose
An affordable open model for agentic coding and terminal automation.
What Z.ai optimised for
Agentic post-training on the existing GLM-5 base.
Unique strengths
- Terminal fluency.
- Open weights.
- Good price.
Where it ranks
Current snapshot · illustrative scores · click any race to watch it
GLM-5.3Best on each axis
Overall
Coding
Hard Reasoning
Math
Science
Writing & Language
Multimodal
Tool Use & Agents
Speed & Cost
Overall Intelligence over time
Best use cases
01
Open DevOps agents
Shell automation, CI fixes and infrastructure scripts.
02
Self-hosted coding assistant
Teams that need to keep code on-prem.
Strengths
Terminal
Among the best open models on our terminal race.
Weaknesses
Breadth
Less capable outside coding.
When to choose it, and when not to
Pick something else when…
- Writing-heavy work.
- Image or document understanding.
Update history
The distilled version first; open any release for the deeper breakdown.
The GLM family