--- title: Everyone's playing 'guess the model' with Ox Alpha description: An unbranded model dubbed Ox Alpha appeared on Open Code, sparking speculation about its maker. It claims a 1 million token context window, multimodal support,... canonical: https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Everyone's playing 'guess the model' with Ox Alpha | daily.dev og:description: An unbranded model dubbed Ox Alpha appeared on Open Code, sparking speculation about its maker. It claims a 1 million token context window, multimodal support,... og:url: https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig og:image: https://api.daily.dev/og/posts/bnd4X6eIg.png og:image:alt: Everyone's playing 'guess the model' with Ox Alpha og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Everyone's playing 'guess the model' with Ox Alpha **[Trends](https://daily.dev/sources/trends)** · 1 min read · 2 upvotes · 0 comments ## Summary An unbranded model dubbed Ox Alpha appeared on Open Code, sparking speculation about its maker. It claims a 1 million token context window, multimodal support, zero data retention, and a claimed 100 trillion tokens/day capacity. It scored 87.5% on the creator's own Kingbench (second behind GLM 5.3, ahead of Opus 4.8 and Qwen 3.8 Max) and outperformed GLM 5.3, Grok 4.6, and GPT 5.6 on a separate 10-task Deep SWA subset. Community fingerprinting of video encoder token counts, tokenizer vocabulary, and writing style points with roughly 90% confidence to Zhipu's next-gen unified multimodal model behind GLM, though nothing is confirmed. Some speculate it may be released as open weight. The free trial is expected to end around August 27th, when the maker may reveal itself. ## Content A free, anonymous model called Ox Alpha showed up on OpenRouter and OpenCode with a 1M-token context window and a claimed 100 trillion tokens/day of capacity, and the internet immediately lost its mind trying to figure out whose servers it's running on. The guessing game has been the actual entertainment here. Scaling01 first floated GLM-5.3 "or something bigger," then walked it back to "Ox Alpha is a ZAI/GLM model." Shreyas_Pandeyy speculated it might be a distilled fusion of GPT-5.6 Sol and Fable, with side bets on SpaceX AI, Kimi Moonshot, MiniMax, or even Sam Altman's rumored Astra project. AleiahLock says it's ## Questions this post answers ### What is the Ox Alpha model and who made it? Ox Alpha is an unbranded AI model that appeared on Open Code without an official maker attached. Community fingerprinting analysis of video encoder token counts, tokenizer vocabulary, writing style, and audio rejection handling points with roughly 90% confidence to it being Zhipu's next-generation unified multimodal model, the company behind GLM, though this remains unconfirmed speculation. _Track unfolding stealth model reveals like this one as they get confirmed, on daily.dev._ ### How does Ox Alpha perform on benchmarks compared to GLM 5.3 and Opus 4.8? Ox Alpha scored 87.5% on Kingbench, placing second behind GLM 5.3 and ahead of Opus 4.8 and Qwen 3.8 Max. On a separate 10-task Deep SWA subset run independently, it outperformed GLM 5.3, Grok 4.6, and GPT 5.6 outright, though the strongest result came from the model creator's own benchmark. _Comparing new model benchmarks against incumbents like GLM and Opus gets easier by following coverage on daily.dev._ ### What specs does the Ox Alpha model claim to offer? Ox Alpha claims a free trial period, a 1 million token context window, multimodal support, zero data retention, and a claimed capacity of 100 trillion tokens processed per day. The free trial is expected to end around August 27th, when the maker may confirm or deny the speculation surrounding its identity. _Developers weighing context window and capacity claims across new models can follow the details on daily.dev._ ## Community take How the wider developer community reacted, aggregated from 1 discussion and 10 comments across x (as of 2026-08-25). **TL;DR:** Reactions center on identifying who actually built Ox Alpha, with mixed reception of its benchmark performance and skepticism about whether the claimed efficiency will hold once pricing is revealed. **Sentiment:** 20% positive · 50% mixed · 30% skeptical **The case for** - Some argue its compute efficiency is underrated and its benchmark score is impressive if achieved cheaply. - One tester's full 113-task run put it near Opus-level performance. **The pushback** - Others note it scores below existing models like 3.7 Flash, undercutting claims of being next-gen. - Several say the real test will be pricing and inference cost, not raw benchmark numbers, since scaffolding could be inflating results. **By community** - x (mixed): Replies mostly speculate about the model's identity and origin, with scattered disagreement over whether its benchmark results are genuinely impressive or just a byproduct of unknown inference costs. **Hottest debate:** Whether Ox Alpha's benchmark scores reflect real capability or are inflated by unclear inference budget and scaffolding. **Open questions** - What pricing will be revealed once the free period ends, and will it change perception of the model's value? - Who actually built Ox Alpha? **Highlights** > @Hesamation @davis7 Mystery resolved? Not from Google then! Can't score below 3.7 Flash to be next gen. > — [ldondeti on x · 1 points](https://x.com/ldondeti/status/2091319599610671567) > @Hesamation @davis7 full 113 task deepswe run came back at 58 percent near opus > — [RaoulDukeDegen on x](https://x.com/RaoulDukeDegen/status/2091314369602953332) > @Hesamation @davis7 The inference budget and scaffolding will decide whether this is a bargain or a benchmark artifact. > — [jurlycat on x](https://x.com/jurlycat/status/2091307689003688027) > @Hesamation @davis7 Ox Alpha's compute efficiency is severely underestimated. > — [QT9277 on x](https://x.com/QT9277/status/2091306161425518801) **Source threads** - [x](https://x.com/Hesamation/status/2091305269170934257) · 0 points · 10 comments --- Tags: [#open-source](https://daily.dev/tags/open-source), [#llm](https://daily.dev/tags/llm) [View this post on daily.dev](https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Everyone's playing 'guess the model' with Ox Alpha","url":"https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig"},"datePublished":"2026-08-21T13:27:16.545Z","dateModified":"2026-08-25T09:22:02.246Z","description":"An unbranded model dubbed Ox Alpha appeared on Open Code, sparking speculation about its maker. It claims a 1 million token context window, multimodal support,...","image":"https://pbs.twimg.com/media/HQP8hOJXwAAVPOJ.jpg","thumbnailUrl":"https://pbs.twimg.com/media/HQP8hOJXwAAVPOJ.jpg","isAccessibleForFree":true,"articleSection":"Trends","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Trends","logo":"https://media.daily.dev/image/upload/s--ZfSp3asX--/f_auto,q_auto/v1780996004/logos/trends?_a=BAMAMiWQ0","url":"https://daily.dev/sources/trends"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"open-source,llm","timeRequired":"PT1M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Trends","item":"https://daily.dev/sources/trends"},{"@type":"ListItem","position":3,"name":"Everyone's playing 'guess the model' with Ox Alpha"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/everyone-s-playing-guess-the-model-with-ox-alpha-bnd4x6eig#faq","mainEntity":[{"@type":"Question","name":"What is the Ox Alpha model and who made it?","acceptedAnswer":{"@type":"Answer","text":"Ox Alpha is an unbranded AI model that appeared on Open Code without an official maker attached. Community fingerprinting analysis of video encoder token counts, tokenizer vocabulary, writing style, and audio rejection handling points with roughly 90% confidence to it being Zhipu's next-generation unified multimodal model, the company behind GLM, though this remains unconfirmed speculation. Track unfolding stealth model reveals like this one as they get confirmed, on daily.dev."}},{"@type":"Question","name":"How does Ox Alpha perform on benchmarks compared to GLM 5.3 and Opus 4.8?","acceptedAnswer":{"@type":"Answer","text":"Ox Alpha scored 87.5% on Kingbench, placing second behind GLM 5.3 and ahead of Opus 4.8 and Qwen 3.8 Max. On a separate 10-task Deep SWA subset run independently, it outperformed GLM 5.3, Grok 4.6, and GPT 5.6 outright, though the strongest result came from the model creator's own benchmark. Comparing new model benchmarks against incumbents like GLM and Opus gets easier by following coverage on daily.dev."}},{"@type":"Question","name":"What specs does the Ox Alpha model claim to offer?","acceptedAnswer":{"@type":"Answer","text":"Ox Alpha claims a free trial period, a 1 million token context window, multimodal support, zero data retention, and a claimed capacity of 100 trillion tokens processed per day. The free trial is expected to end around August 27th, when the maker may confirm or deny the speculation surrounding its identity. Developers weighing context window and capacity claims across new models can follow the details on daily.dev."}}]} ```