GLM-5.1 versus GLM-5.3

What are we comparing?

GLM-5.1 and GLM-5.3 are two open-weight models that you can install on your own servers, both built by Zhipu AI (Z.ai). GLM-5.1 was released on April 7, 2026, GLM-5.3 on August 14, 2026.

All figures below come from public sources checked on October 22, 2026: models.dev and OpenRouter for prices and capabilities, Epoch AI for scores.

Which one is more capable?

On our capability index, GLM-5.1 scores 70.9 out of 100 and GLM-5.3 78.8. GLM-5.3 leads by 7.9 points.

On the 4 benchmarks reported for both, GLM-5.1 wins 1 time and GLM-5.3 3:

• GPQA Diamond (PhD-level science questions): GLM-5.1 89.9%, GLM-5.3 90.9%
• OTIS Mock AIME (competition mathematics problems): GLM-5.1 93.3%, GLM-5.3 91.1%
• FrontierMath (very hard research-level mathematics): GLM-5.1 36.8%, GLM-5.3 68.8%
• APEX-Agents (professional tasks carried out by an agent): GLM-5.1 40.9%, GLM-5.3 56.6%

Pricing

GLM-5.1 costs $0.90 per million input tokens and $3.59 per million output tokens; GLM-5.3 costs $1.2 and $4.18.

For a typical agent task (read 20,000 tokens, write 2,000), expect $0.025 with GLM-5.1 versus $0.032 with GLM-5.3. GLM-5.1 is 1.3 times cheaper, a gap that adds up quickly over 1,000 tasks: $7.17 difference.

Run the numbers with your own volumes in the agent cost calculator.

Under the hood

GLM-5.3 reads more in one go: 1,000,000 tokens of context (roughly 1,500 pages), versus 200,000 for GLM-5.1.

Reasoning mode: GLM-5.1 yes, GLM-5.3 yes. Tool calling: GLM-5.1 yes, GLM-5.3 yes. Image input: GLM-5.1 no, GLM-5.3 no.

Both publish their weights: you can host them yourself.

In short

Our reading of the data:

• Choose GLM-5.1 for a lower cost.
• Choose GLM-5.3 for a higher capability index, a larger context window and a more recent release.

This verdict relies on public figures; it does not replace a trial on your own tasks.