--- format: "aidr-story-markdown/v1" id: "87bcf2981065f7af226962e645f9330df0121a9c393f20e2a8dc1ebfb2740eca" canonical_url: "https://aidr.today/87bcf298?lang=en" title: "Z.ai Open Sources Ox Alpha GLM Weights Tonight After OpenRouter Sweep" lang: "en" requested_lang: "en" available_langs: ["en","vi"] translation_fallback: null fallback_fields: [] published_at: "2026-08-26T11:55:39.000Z" category: "Releases" topics: ["alibaba","qwen","open-source","agent","inference"] source_urls: ["https://huggingnews.com/ai/update-zai-open-sources-ox-alpha-glm-weights-tonight-after-openrouter-sw-1c6a0494","https://x.com/Techmeme/status/2092548255464960235","https://x.com/kimmonismus/status/2092561180128608672","https://x.com/MaxForAI/status/2092554532614271415","https://x.com/Hesamation/status/2092565564342341805","https://x.com/vipulved/status/2092387838566482181","https://x.com/AAAzzam/status/2092391013595677140","https://x.com/togethercompute/status/2092388843429478751"] summary: "The AI model Ox Alpha surged to the top of usage charts before its developer Z.ai revealed the tool is a new iteration of the GLM series. The company will release the open weights tonight, resolving speculation over the model's origins. The tool, identified as GLM-5.3 Flash, was developed by the entity also known as Zhipu. The Flash model processed 11.6 trillion tokens in three days, 2.6 times the scale of the previous biggest model launch on OpenRouter. Z.ai equipped the model with a 1.05 million token context window to support coding agents and sustained agentic work, prioritizing low cost and high speed over the previous GLM-5.3 SOTA release." --- # Z\.ai Open Sources Ox Alpha GLM Weights Tonight After OpenRouter Sweep > [Open the canonical story]() **Published:** 2026-08-26T11:55:39.000Z **Category:** Releases **Topics:** alibaba, qwen, open\-source, agent, inference ## Summary The AI model Ox Alpha surged to the top of usage charts before its developer Z\.ai revealed the tool is a new iteration of the GLM series\. The company will release the open weights tonight, resolving speculation over the model's origins\. The tool, identified as GLM\-5\.3 Flash, was developed by the entity also known as Zhipu\. The Flash model processed 11\.6 trillion tokens in three days, 2\.6 times the scale of the previous biggest model launch on OpenRouter\. Z\.ai equipped the model with a 1\.05 million token context window to support coding agents and sustained agentic work, prioritizing low cost and high speed over the previous GLM\-5\.3 SOTA release\. ## Sources - [Story source]() - [Story source]() - [Supporting source]() - [Supporting source]() - [Supporting source]() - [Story source]() - [Story source]() - [Supporting source]()