Z AI GLM 5.3 Flash (Ox Alpha) Release
5 videos · 5 channels · score 131k
Zhipu AI released the open-weight GLM 5.3 Flash model, appearing as 'Ox Alpha' with MIT-licensed weights, which creators praise for delivering near-frontier multimodal performance at a fraction of the cost of competitors like Claude, though some note issues with speed and looping behavior.
View volume · log scale
The coverage — 5 videos
The mystery is solved... and the answer is 40x cheaper than Claude
Zhipu AI's anonymously released 'Ox Alpha' model on Open Router was revealed as GLM 5.3 Flash with MIT-licensed weights, priced 40x cheaper than Claude but criticized for being slow and loop-prone, with a 50% discount through September 9th.
GLM 5.3: Powerful AI Is Becoming Almost Free
GLM 5.3 and its stealth-released Flash variant, using linear attention and index pool techniques, overtook DeepSeek in usage and prompted the creator to argue that such efficiency innovations could make powerful AI nearly free and surpass Fable within months.
Cancel your subscriptions, Ox-Alpha is here! (GLM 5.3 Flash)
Z AI secretly released GLM 5.3 Flash as 'Ox Alpha' on Open Router, a 320B-param open-weights MoE model delivering near-frontier performance at 2-3% of Claude Fable 5's cost, which the creator calls a game-changer for affordability and control.
Ox Alpha Is INSANE
Ox Alpha, a flash version of GLM 5.3 with full multimodality on Chinese AI chips, launches as the creator's most-used model, matching Opus 4.8 at a fraction of the cost.
All Roads Lead back To GLM!
Zhipu AI released GLM-53 and the smaller 320B-parameter GLM-53 Flash vision-language model, which the creator now favors over DeepSeek V4 Flash for its superior vision and faster time-to-fail behavior, avoiding hallucination-driven wasted time.