Mystery Model Ox Alpha Revealed as Zhipu's GLM-5.3-Flash, Served on Domestic Chinese Chips

The mysterious model Ox Alpha, which drew speculation that its massive OpenRouter traffic was Gemini, has been confirmed by Bloomberg and multiple researchers as Zhipu's GLM-5.3-Flash (320B-A18B). After the free trial ended, it officially launched on OpenRouter via Novita. Its record-breaking inference traffic was reportedly served entirely on domestic Chinese chips at roughly 1% of frontier-model pricing, directly disrupting the cost-efficient inference market.

Confirmed

• OpenRouter announced the end of the Ox Alpha free trial, succeeded by GLM-5.3 Flash (per Teknium and paulrichnb); Ox Alpha had processed over 20 trillion tokens in 6 days. TheTuringPost reported 8 million sessions and 24 trillion tokens of real-world testing in 4 days. jietang noted an LMSys Arena score of AA=57 and 20% of OpenRouter's weekly token share, ranking first.
• SimonOlsn showed Ox Alpha and GLM-5.3-Flash have identical weights with only a different name, exposing model re-skinning in the API market; eyishazyer noted a researcher pieced together clues from tokenizer leaks and error codes over six days, with weights released the night Bloomberg confirmed.
• The model comes from Zaiorg: the first natively multimodal GLM-5 model, 320B total parameters with 18B active; a 1M-token context window; priced at one-tenth of GLM-5.2 with half the layers; OpenAI-compatible API with tool calling and Anthropic-format support (AccBalanced, rohanpaulai).
• pstAsiatech reported the stealth test ran on a 100,000-chip domestic cluster with record usage within three days of launch; bindureddy detailed Zhipu's self-developed SGLang inference engine; teortaxesTex noted the chips are not on the most advanced process node.
• YuchenjUW and andrewncarr noted earlier Gemini speculation was disproven, confirming domestic accelerators rather than NVIDIA, echoing Jensen Huang's view that export controls accelerate China's independent compute stack.

Unconfirmed

• The compute provider remains disputed: bindureddy early on speculated Nvidia supplied capacity for the massive free quota, while AccBalanced suggested Huawei instead; Zhipu's official line is "domestic chips" without naming a vendor.
• Cumulative token counts vary across sources: OpenRouter says 20+ trillion in 6 days, bindureddy says 42 trillion, pstAsiatech says 62 trillion cumulative with up to 100 trillion tokens daily—figures await verification.

Why it matters

• Frontier-adjacent Arena performance at 1% of the price, served on purely domestic Chinese chips at this scale, directly disrupts the cost-efficient inference market and validates Jensen Huang's warning that export controls are accelerating China's independent compute stack.
• The Ox Alpha re-skinning case exposes transparency gaps in anonymous model releases and platform disclosure, providing a canonical example for API market governance.

2026-08-26 ~ 2026-08-28 · 19 related posts

Full story(16 episodes)→

• Episode 1: Mystery model Ox Alpha stuns community; fingerprints point to Zhipu GLM-5.3(2026-08-21, 57 posts)
• Episode 2: OpenRouter's Mystery 'Ox Alpha' Model Identified as GLM(2026-08-21, 3 posts)
• Episode 3: Cola Launches Ox Model Claiming Terra-Level Capability at Flash Speed(2026-08-21, 2 posts)
• Episode 4: Mysterious Model Ox Alpha Gives Away 100T Tokens Daily, Sparking Power Questions(2026-08-21, 2 posts)
• Episode 5: Ox-Alpha Underperforms in Tests, Called Marketing Stunt(2026-08-22, 3 posts)
• Episode 6: Ox Alpha scores ~63% on full DeepSWE; earlier 80% was subset artifact(2026-08-22, 6 posts)
• Episode 7: Forensics Point to Zhipu's Unreleased GLM Behind Mystery Model Ox Alpha(2026-08-22, 31 posts)
• Episode 8: Ox Alpha Generates 64K-Token 3D Scene in One Prompt(2026-08-22, 2 posts)
• Episode 9: Mystery model Ox Alpha draws mixed reviews from early testers(2026-08-22, 6 posts)
• Episode 10: Mystery Model Ox Alpha Launches Free, Identity Unknown(2026-08-23, 3 posts)
• Episode 11: Mysterious Model Ox Alpha Debuts on OpenRouter, Sparking Speculation(2026-08-24, 5 posts)
• Episode 12: Leak: Ox Alpha Is GLM-5.3 Flash, 320B MoE at One-Tenth the Price(2026-08-25, 4 posts)
• Episode 13: Mystery Model Ox Alpha Revealed as Zhipu's GLM-5.3 Flash, Weights Going Open-Source Tonight(2026-08-25, 17 posts)
• Episode 14: Stealth Model Ox Alpha Proves Mediocre at Coding, Trails Grok 4.6(2026-08-25, 3 posts)
• Episode 15: Mystery Model Ox Alpha Revealed as Zhipu's GLM-5.3-Flash, Served on Domestic Chinese Chips(2026-08-26, 19 posts)
• Episode 16: Mystery Model Ox Alpha Burns 42T Tokens in Six Days(2026-08-26, 3 posts)

• Nvidia may have funded 100T tokens for free GLM 5.3 Flash release — bindureddy · 2026-08-26
• Speculation: Nvidia Powered GLM's 100T Token Free Giveaway — bindureddy · 2026-08-26
• [source] GLM-5.3 Flash Arrives on OpenRouter, Succeeding Ox Alpha — Teknium · 2026-08-26
• GLM-5.3-Flash handles 100T tokens daily, running entirely on Chinese chips — airesearch12 · 2026-08-26
• ZAI Launches GLM-5.3 Flash Model on OpenRouter — paulrich_nb · 2026-08-27
• Rumor: Huawei, Not Nvidia, Backed GLM 5.3 Flash's Massive Token Run — AccBalanced · 2026-08-27
• Zhipu GLM-5.3-Flash launches on OpenRouter with 1M-token context — AccBalanced · 2026-08-27
• [source] Zhipu Reveals Ox Alpha as GLM-5.3-Flash, Serving 100T Tokens/Day on Chinese GPUs — pstAsiatech · 2026-08-27
• Zhipu's GLM 3.5 Flash Served 42T Tokens in 6 Days Free on Chinese Chips — bindureddy · 2026-08-27
• GLM-5.3-Flash Hits 100T Tokens/Day on Domestic Chips — Yuchenj_UW · 2026-08-27
• 100T tokens/day traced to GLM-5.3-Flash on Chinese chips, validating export control impact — andrew_n_carr · 2026-08-27
• GLM-5.3 Flash tops OpenRouter share with 1% frontier cost on Chinese chips — jietang · 2026-08-27
• Ox Alpha Revealed as Zhipu's GLM-5.3-Flash, Running on Chinese Chips All Along — eyishazyer · 2026-08-27
• Zhipu's GLM-5.3-Flash Served Record Traffic Entirely on Domestic Chinese Chips — teortaxesTex · 2026-08-27
• [source] OpenRouter's 'Ox Alpha' Revealed as Zhipu GLM-5.3-Flash Repackaging — SimonOlsn · 2026-08-27
• Ox Alpha revealed to be a rebrand of Zhipu's GLM-5.3-Flash — SimonOlsn · 2026-08-27
• Z.ai reveals Ox Alpha is GLM-5.3-Flash running on domestic chips — rohanpaul_ai · 2026-08-27
• Zhipu's GLM-5.3 Runs Entirely on Domestic Chips, Processing 62 Trillion Tokens — pstAsiatech · 2026-08-27
• Anonymous Ox Alpha revealed as Zhipu's GLM-5.3-Flash after massive stealth test — TheTuringPost · 2026-08-28