Ox Alpha – A mysterious new AI model

Ox Alpha, a new reasoning-first AI model with a million-token context window, has appeared without an identified creator, and an independent community benchmark of 10 real-world coding tasks shows it solved 8 (80%), ahead of reference models fable-5, glm-5.3, gpt-5.6-sol, and grok-4.6, which scored 65%, 62%, 52%, and 62% respectively. The model is designed for long-horizon coding, supports text, image, and video input, native tool calling, structured output, and is available with no account needed.

Meet Ox Alpha. The reasoning model that appeared out of nowhere. Built for code, long-horizon agents, and a million tokens of context. Nobody knows who made it — everyone wants to try it. text · image · video Thinks before it speaks. Ox Alpha is a reasoning-first model. It plans, checks itself, and only then answers — so hard problems get worked through instead of guessed at. - Multi-step logic Math, systems design, proofs, tricky edge cases — worked through, not pattern-matched. - Visible thinking Watch the reasoning stream in before the final answer lands. - Production-grade output Tuned for real workloads, not party tricks. A million tokens of memory. Paste an entire codebase, a 500-page spec, or a week of transcripts. Ox Alpha keeps all of it in mind while it works — no chunking, no retrieval tricks, no losing the thread. Built for real work. Everything a frontier model should do — and a few things most of them can't. Long-horizon coding Designed for sustained software engineering: debugging, refactoring, and reasoning across large codebases without losing context. Sees what you see Text, images, and video in. Screenshots, diagrams, and UI — reasoned about, not just described. Tool calling Native tools and tool choice support for agents that actually do things. Structured output Ask for JSON and get JSON — schema-shaped responses via response format . Sustained agentic work Keeps its footing across long, multi-step tasks instead of drifting off halfway. Watch it reason. A reasoning model shows its work. Here's what a typical exchange looks like. cache = {}; def get k : return cache.get k or fetch k Thinking Two issues: or treats falsy cached values 0, "", as misses and re-fetches; and nothing ever writes to cache , so the "stale" data must come from fetch itself — likely a module-level connection reused across requests. Check both. or : any falsy cached value is treated as a miss. Also, cache is never populated. Use an explicit membership check and store results: python def get k : if k not in cache: explicit miss check cache k = fetch k actually populate return cache k If results still look stale, the culprit is fetch reusing a connection with read-consistency lag — want me to look at it? Stealth model. Top of the table. On an independent run of 10 real-world coding tasks, Ox Alpha solved 8 — ahead of every frontier model it was lined up against. | Task | fable-5 max | glm-5.3 max | gpt-5.6-sol max | grok-4.6 xhigh | ox-alpha | |---|---|---|---|---|---| | anko-typed-variable-bindings | 4/4 | 4/4 | 2/4 | 1/4 | ✓ | | arktype-json-schema-refs | 2/4 | 1/4 | 3/4 | 1/4 | ✓ | | fastapi-deprecation-headers | 4/4 | 3/4 | 3/4 | 4/4 | ✓ | | helm-unified-manifest-stream | 4/4 | 4/4 | 4/4 | 4/4 | ✓ | | igel-persist-feature-schema | 3/4 | 3/4 | 0/4 | 4/4 | ✓ | | katex-multicolumn-array-spans | 2/4 | 4/4 | 3/4 | 4/4 | ✓ | | meriyah-explicit-resource-decl | 1/4 | 0/4 | 0/4 | 0/4 | ✓ | | query-persist-restored-state | 2/4 | 3/4 | 1/4 | 2/4 | ✓ | | scc-bounded-memory-spilling | 4/4 | 3/4 | 4/4 | 4/4 | ✕ | | vulture-persistent-analysis-cache | 0/4 | 0/4 | 1/4 | 1/4 | ✕ | | Mean on these 10 | 65% | 62% | 52% | 62% | 80% | Independent community benchmark — 10 real-world coding tasks. Reference models: passes out of 4 attempts per task. Ox Alpha: pass/fail. Highlighted row: the task every reference model scored 1/4 or worse on — Ox Alpha solved it. Third-party data, small sample — directional, not definitive. Pick a task. Or bring your own. Zero to answer in seconds. Get started in seconds — no account needed. Describe the task Type a goal in plain English. Paste the whole file, log, or document — context is not a problem. Ox Alpha reasons It thinks through the problem step by step, then streams back its answer in real time. Review & iterate Refine with follow-ups. The whole conversation stays in its 1M-token memory. Questions, answered. What is Ox Alpha? What is oxalpha.com? Is Ox Alpha really free? Who made Ox Alpha? What is Ox Alpha good at? Is my conversation data private? Can I use Ox Alpha on mobile devices? Still curious? Get in touch /contact . oxalpha.com is a fast, free way to chat with Ox Alpha — the stealth reasoning model that appeared in August 2026 — right from your browser. No account needed, no subscriptions, no barriers. Ox Alpha is built for serious work: long-horizon software engineering, complex multi-step reasoning, and tasks that mix text with visual context. With a 1,048,576-token context window, it can hold entire codebases, papers, or transcripts in mind while it works. The interface is clean and simple. Type a question or prompt, and Ox Alpha reasons through it and responds in real time. Keep asking follow-ups, start new topics, or stress-test it — that's exactly what a stealth period is for. oxalpha.com is an independent project and is not affiliated with the unknown lab behind the model. No personal information is required, and we don't store your conversations on our servers. Try it before the internet figures out what it is. Free while Ox Alpha stays in stealth. No account, no card, no waitlist. Open chat /chat