Zhipu’s GLM models are coding-first open-weight flagships: GLM-5.3 posts a 50% coding jump over 5.2 and near-Mythos cyber capability, while GLM-5.3 Flash brings multimodal speed at MIT-licensed prices.
GLM-5.3 API live 18 Aug 2026; open weights 28 Aug under the GLM-5.3 licence.
Who they are
Zhipu AI (Z.ai, HKEX-listed) is China’s coding-model specialist. GLM-5.3 trains on a ~743B base with ~1M context; GLM-5.3 Flash (the former stealth “Ox Alpha”) runs 320B/18B active multimodal, served on 100k domestic chips.
320B/18B multimodal MoE at 1.05M ctx — the speed tier
US$0.15 / US$0.50 per 1M (≈ RM0.61 / RM2.04)
GLM-5.2previous gen · MIT
Still available at the same rates as 5.3
US$1.40 / US$4.40 per 1M (≈ RM5.71 / RM17.95)
Last updated — Vendor list prices (Sep 2026), approx RM at 1 USD ≈ RM4.08 (spot, 19 Sep 2026). GLM-5.3 open weights carry a custom licence (firms above US$10B revenue need security review). On BD LLM Token Hub today. Figures are vendor public list prices in USD per 1M tokens, standard tier — no batch, cache, promo or peak-hour discount applied. Vendors change list prices without notice: confirm on the vendor’s own pricing page before budgeting.
Historical models — for reference
Discontinued or fully superseded models, kept for historical reference. Active models — even previous-generation ones still on sale — always stay in the lineup above.
Model
Era & what it was for
Historic list price
GLM-52026
Previous flagship generation
US$1.00 / US$3.20 per 1M (≈ RM4.08 / RM13.06) — historic list
GLM-4.72025
The 4.x-era flagship before the 5 line
US$0.60 / US$2.20 per 1M (≈ RM2.45 / RM8.98) — historic list
Latest news
Recent announcements with outlet links (EN first, CN where available).
28 Aug 2026
GLM-5.3 open weights — and a global cybersecurity debate
Zhipu released 5.3 weights under a novel licence; the NYT framed it as a test of world cybersecurity, citing its Mythos-5-level defensive capability.