Z.ai releases GLM-5.3-Flash, the first natively multimodal GLM-5 series model, with 320B parameters, saying it outperforms GLM-5.2 at "one-tenth the price" (Z.ai)
Z.ai says GLM-5.3-Flash was the previously anonymous Ox Alpha model, now released with open weights.
The model is a 320B-parameter MoE with 18B active parameters and native multimodal support, according to Z.ai’s post. Techmeme’s roundup notes Z.ai’s claim that it beats GLM-5.2 at one-tenth the price, with API pricing listed at $0.15 per million input tokens and $0.50 per million output tokens. The item also points to reports that Ox Alpha ran through OpenCode and OpenRouter before the reveal, with traffic served on Chinese AI chips. Techmeme's note
The model is a 320B-parameter MoE with 18B active parameters and native multimodal support, according to Z.ai’s post. Techmeme’s roundup notes Z.ai’s claim that it beats GLM-5.2 at one-tenth the price, with API pricing listed at $0.15 per million input tokens and $0.50 per million output tokens. The item also points to reports that Ox Alpha ran through OpenCode and OpenRouter before the reveal, with traffic served on Chinese AI chips. Techmeme's note
score 7