Back
All GLM models
LLMS • CURATED • UPDATED AUG 14, 2026

GLM 5.3

Z.ai's post-trained coding and agentic model on the GLM-5.2 base

GLM-5.3 is Z.ai's flagship coding and agentic model, released 14 August 2026. It uses the same 743B-parameter mixture-of-experts base as GLM-5.2, with Z.ai attributing the gains to scaled post-training rather than a new pre-training run. It targets long-horizon agentic coding, business-process automation, defensive security work and tasks that span many steps. The model supports three reasoning-effort levels and a 1M-token route for coding plans.

Pricing Paid
Platforms web, api
API Yes
Open Source No
Modalities LLMs, IDEs & Coding Tools
Best For Best for Post-Training Gains
Date Added 2026-08-14

The Terminal-Bench 3.0 score moved from 4.6% to 28.3%, DeepSWE v1.1 from 46.2% to 66.9%, and CyberGym from 77.2% to 84.5% — all on the same base as GLM-5.2. That is a real signal about post-training returns, even if most numbers are vendor-run and weights are not yet released. It is also reported as one of the fastest models in its class, at roughly 115 tokens per second.

Call GLM-5.3 through the Z.ai API or the chat interface at chat.z.ai. Set `thinking.type` to `enabled` and use `reasoning_effort` of `low`, `high` or `max` (default). For the 1M coding-plan route, use the `glm-5.3[1m]` suffix. Check the Z.ai pricing page for current rates before running large workloads.

Terminal-Bench 3.0 · ide-coding
28.3%
DeepSWE v1.1 · ide-coding
66.9%
CyberGym · llm
84.5%
1 Change `thinking.type: disabled` to `enabled` with `reasoning_effort: low` before switching model IDs
2 Benchmark with the same agent harness and time budget as Kimi K3 and Qwen3.8-Max
3 Use `max` effort for long-horizon coding, `low` for latency-sensitive paths
4 Wait for the open-weight release before planning self-hosted deployment
5 Confirm the standard API price, not the GLM-5.2 price, before budgeting
Website API Docs Pricing
Google Antigravity 2.0 Claude Fable 5 NotebookLM Claude Opus 5 GPT-5.6 Sol

Long-Horizon Refactor

Run a multi-file refactoring task that spans many tool calls.

STEPS:
  1. Describe the refactor in a single detailed prompt
  2. Set `reasoning_effort` to `max`
  3. Give the agent access to the relevant files and tests
  4. Review every file change before applying
  5. Measure wall-clock time and output tokens against GLM-5.2
Paid

Requires a paid subscription.

View pricing details →
📚

CLAUDE.md In Practice: What Actually Earns A Place In The File

The most-starred CLAUDE.md on GitHub has four rules, not twelve, and the error rates everyone quotes...

Open Models: What They Cannot Do

The benchmark gap has largely closed: open models now score within a point of the closed frontier on...

Hardware For Local AI Models: What The Memory Shortage Changed

Capacity decides whether a model runs. Bandwidth decides how fast it talks, and it is the number mos...

How To Run AI Models Locally: Hardware, Quantisation And Engines

Running a frontier open model locally is a memory problem before it is a speed problem, and a bandwi...

Uncensored AI Models: What Abliteration Actually Does To Them

Abliteration removes a model's ability to refuse by deleting one direction from its activations. It ...

View GLM 5.3 Alternatives (2026) →

Compare GLM 5.3 with 5+ similar llms AI tools.

Q

Is GLM 5.3 free?

A

No, GLM 5.3 requires a paid subscription.

Q

Does GLM 5.3 have an API?

A

Yes, GLM 5.3 offers an API for programmatic integration.

Q

What is GLM 5.3 best for?

A

GLM 5.3 is best for Post-Training Gains. GLM-5. The Terminal-Bench 3.0 score moved from 4.6% to 28.3%, DeepSWE v1.1 from 46.2% to 66.9%, and CyberGym from 77.2% to 84.5% — all on the same base as GLM-5.2. That is a real signal about post-training returns, even if most numbers are vendor-run and weights are not yet released. It is also reported as one of the fastest models in its class, at roughly 115 tokens per second.

Q

What platforms does GLM 5.3 support?

A

GLM 5.3 supports web, api.

Q

Is GLM 5.3 open source?

A

No, GLM 5.3 is not open source.

Q

How do I get started with GLM 5.3?

A

Call GLM-5.3 through the Z.ai API or the chat interface at chat.z.ai. Set `thinking.type` to `enabled` and use `reasoning_effort` of `low`, `high` or `max` (default). For the 1M coding-plan route, use the `glm-5.3[1m]` suffix. Check the Z.ai pricing page for current rates before running large worklo...

Q

How do I use GLM 5.3?

A

GLM 5.3 is a large language model for text generation, analysis, and conversation. Access through the web interface. Enter prompts or questions to get responses. It excels at terminal-bench 3.0 jumped from 4.6% to 28.3% on the same base.

🏷️

Work on GLM 5.3? You're hand-reviewed in our directory. Add this badge to your site — it links back to this profile.

Featured on CuratedAI