Zhipu's GLM-5.3-Highspeed Appears in ZCode Gray Release, With Users Reporting About 490 Tokens/s (2 minute read)
Summary
Zhipu's GLM-5.3-Highspeed model has appeared in a ZCode gray release, with early users reporting inference speeds of around 490 tokens/s. The model is being rolled out gradually to randomly selected subscribers.
Similar Articles
@Yuchenj_UW: GLM-5.3 at 310 tok/s! Databricks inference is #1 in both speed and latency, again. On our internal Databricks coding be…
GLM-5.3 achieves 310 tokens per second on Databricks inference, leading in both speed and latency, and is the strongest open-source model for coding, competitive with Fable 5 and Opus 4.8.
@TechMDAI: GLM-5.3-Flash EXL3-3.0bpw by @0xSero @BrandonMusicKy @LottoLabs @localmaxxing 193.8 tk/s
The article highlights the GLM-5.3-Flash EXL3-3.0bpw AI model with an inference speed of 193.8 tokens per second, attributed to multiple contributors.
zai-org/GLM-5.2-FP8
Z.AI releases GLM-5.2, a flagship open-source model with a solid 1M-token context, improved coding capabilities, and a new IndexShare sparse attention architecture that reduces FLOPs by 2.9x at 1M context.
I did some model hacks, and got GLM5.2 from about 2.5 tok/s to >50 tok/s on my GH200 system.
A detailed blog post describing how to dramatically speed up GLM-5.2 inference on a dual Grace Hopper system from 2.5 tok/s to over 50 tok/s by stopping model cross-module traffic and grafting an FP8 MTP head onto the INT4 base.
GLM-5.3 hits the API at 1.4/4.4 per million tokens (2 minute read)
GLM-5.3, a new frontier open-source language model from Chinese startup Z.ai, has launched its API at $1.40 per million input tokens and $1.40 per million input tokens and $4.40 per million output tokens, with plans for open-weight release. The model offers advanced coding capabilities and competitive pricing compared to other leading AI models.