FULL STORY
GLM-5.3-Flash: From Discount to Open Source
Zhipu's GLM-5.3-Flash launched with a limited-time discount on Merge Gateway, touting coding costs 26x cheaper than GPT-5, before Z.ai officially open-sourced the 320B, million-context model.
2026-08-30 ~ 2026-09-03 · 2 episodes · 4 posts
Episode 1 · GLM 5.3 Flash Discounted 10%, 26x Cheaper Than GPT-5 for Coding (2026-08-30, 2 posts)
Merge Gateway is offering a limited 10% discount on GLM 5.3 Flash through the end of September. The model is already roughly 26x cheaper than GPT-5 for coding tasks ($35 vs $975 per billion tokens) while outscoring it on the Arena coding leaderboard.
- GLM-5.3-Flash Beats GPT-5 in Coding Arena at 26x Lower Cost — ccerrato147 · 2026-08-30
- GLM 5.3 Flash Discounted 90%, Cuts Task Costs Significantly — shensi · 2026-09-01
Episode 2 · Z.ai Open-Sources GLM-5.3-Flash with 320B Params and 1M Context (2026-09-01, 2 posts)
Zhipu (Z.ai) has open-sourced GLM-5.3-Flash (codename "Ox Alpha"), a 320B/18B-activated MoE model with a 1M-token context window, multimodal support, and NVFP4 quantization, priced at about one-tenth of comparable models.
- Z.ai Releases GLM-5.3-Flash: 320B Params, 1M Context, and NVFP4 Quantization — alejandroll10 · 2026-09-01
- Zhipu open-sources GLM-5.3-Flash: a 320B/18B MoE with hybrid attention and 1M-token context at 1/10 the price — dl_weekly · 2026-09-03