Ordinal

Which model is worth using

1 Oct 2026

Z.ai

GLM-5.3 Flash

The cost-frontier open model. A serious score, a serious license, and a price that makes a routing layer obvious.

Open weights
42Index · Published
$0.25Per measured task
168 pts/$Points / $
45Tokens / sec

No High score is published. Showing Published.

What this workload costs

List price for one call. Reasoning tokens count as output. A cache hit uses the lab’s cache rate when one is published.

GLM-5.3 Flash
Cache applied
$0.0040

Specs

API id
Not published here
Released
Sep 2026
Context
1M
Max output
128k
Input / 1M
$0.15
Output / 1M
$0.50
Cache read / 1M
$0.03
Knowledge
—
Inputs
Text, Image, Video
Weights
MIT

List price after the launch promo ended on 9 Sep 2026. The promo was $0.075 / $0.25. Some hosts still sell below list. About 320B total, 18B active.

Text, image, video, and file input, from the 3 Sep 2026 vendor-price table.

Use it when

The open model you put on the cheap path. Index 42, multimodal input, MIT weights, $0.15 / $0.50 list, and $0.25 per measured task.

Leave it when

The last few points of frontier work. Sol, Grok, and MiMo-V2.6-Pro sit above it.

Sources