Chinese startup z.ai has opened API access to its new open-source language model GLM-5.3, letting developers integrate it into agents and applications at the same price as its predecessor. The model charges $1.40 per million input tokens and $4.40 per million output tokens, matching GLM-5.2’s rates. Cached inputs cost $0.26 per million tokens, with storage currently free for a limited time.
GLM-5.3’s API debut follows reports of advanced cyber capabilities, including uncovering an undetected vulnerability in Cursor. Independent benchmarks from Artificial Analysis rank it at 60 on the Intelligence Index—tied for the top open-weights model and seven points ahead of GLM-5.2—though its higher verbosity means per-task costs may rise despite flat token pricing.
At $5.80 per million input and output tokens, GLM-5.3 undercuts several premium models like Grok 4.6 and Kimi K3. Developers already subscribed to z.ai’s GLM Coding Plan can call the new model through the OpenAI Chat Completions-compatible protocol, while z.ai plans to release the model’s weights publicly, though no date or licensing terms are set.
