{"slug": "ox-alpha-a-mysterious-new-ai-model", "title": "Ox Alpha – A mysterious new AI model", "summary": "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.", "body_md": "# Meet Ox Alpha.\n\nThe 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.\n\ntext · image · video\n\n## Thinks before\n\nit speaks.\n\nOx Alpha is a reasoning-first model. It plans, checks itself, and only then answers — so hard problems get worked through instead of guessed at.\n\n-\n**Multi-step logic** Math, systems design, proofs, tricky edge cases — worked through, not pattern-matched.\n\n-\n**Visible thinking** Watch the reasoning stream in before the final answer lands.\n\n-\n**Production-grade output** Tuned for real workloads, not party tricks.\n\n## A million tokens of memory.\n\nPaste 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.\n\n## Built for real work.\n\nEverything a frontier model should do — and a few things most of them can't.\n\n### Long-horizon coding\n\nDesigned for sustained software engineering: debugging, refactoring, and reasoning across large codebases without losing context.\n\n### Sees what you see\n\nText, images, and video in. Screenshots, diagrams, and UI — reasoned about, not just described.\n\n### Tool calling\n\nNative `tools`\n\nand `tool_choice`\n\nsupport for agents that actually do things.\n\n### Structured output\n\nAsk for JSON and get JSON — schema-shaped responses via `response_format`\n\n.\n\n### Sustained agentic work\n\nKeeps its footing across long, multi-step tasks instead of drifting off halfway.\n\n## Watch it reason.\n\nA reasoning model shows its work. Here's what a typical exchange looks like.\n\n`cache = {}; def get(k): return cache.get(k) or fetch(k)`\n\n**Thinking** Two issues:\n\n`or`\n\ntreats falsy cached values (0, \"\", []) as misses and re-fetches; and nothing ever writes to `cache`\n\n, so the \"stale\" data must come from `fetch`\n\nitself — likely a module-level connection reused across requests. Check both.**or**: any falsy cached value is treated as a miss. Also,\n\n`cache`\n\nis never populated. Use an explicit membership check and store results:\n\n``` python\ndef get(k):\n    if k not in cache:      # explicit miss check\n        cache[k] = fetch(k)   # actually populate\n    return cache[k]\n```\n\nIf results still look stale, the culprit is`fetch`\n\nreusing a connection with read-consistency lag — want me to look at it?\n## Stealth model. Top of the table.\n\nOn an independent run of 10 real-world coding tasks, Ox Alpha solved 8 — ahead of every frontier model it was lined up against.\n\n| Task | fable-5 [max] | glm-5.3 [max] | gpt-5.6-sol [max] | grok-4.6 [xhigh] | ox-alpha |\n|---|---|---|---|---|---|\n| anko-typed-variable-bindings | 4/4 | 4/4 | 2/4 | 1/4 | ✓ |\n| arktype-json-schema-refs | 2/4 | 1/4 | 3/4 | 1/4 | ✓ |\n| fastapi-deprecation-headers | 4/4 | 3/4 | 3/4 | 4/4 | ✓ |\n| helm-unified-manifest-stream | 4/4 | 4/4 | 4/4 | 4/4 | ✓ |\n| igel-persist-feature-schema | 3/4 | 3/4 | 0/4 | 4/4 | ✓ |\n| katex-multicolumn-array-spans | 2/4 | 4/4 | 3/4 | 4/4 | ✓ |\n| meriyah-explicit-resource-decl | 1/4 | 0/4 | 0/4 | 0/4 | ✓ |\n| query-persist-restored-state | 2/4 | 3/4 | 1/4 | 2/4 | ✓ |\n| scc-bounded-memory-spilling | 4/4 | 3/4 | 4/4 | 4/4 | ✕ |\n| vulture-persistent-analysis-cache | 0/4 | 0/4 | 1/4 | 1/4 | ✕ |\n| Mean on these 10 | 65% | 62% | 52% | 62% | 80% |\n\nIndependent 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.\n\n## Pick a task. Or bring your own.\n\n## Zero to answer in seconds.\n\nGet started in seconds — no account needed.\n\n### Describe the task\n\nType a goal in plain English. Paste the whole file, log, or document — context is not a problem.\n\n### Ox Alpha reasons\n\nIt thinks through the problem step by step, then streams back its answer in real time.\n\n### Review & iterate\n\nRefine with follow-ups. The whole conversation stays in its 1M-token memory.\n\n## Questions, answered.\n\n## What is Ox Alpha?\n\n## What is oxalpha.com?\n\n## Is Ox Alpha really free?\n\n## Who made Ox Alpha?\n\n## What is Ox Alpha good at?\n\n## Is my conversation data private?\n\n## Can I use Ox Alpha on mobile devices?\n\nStill curious? [Get in touch](/contact).\n\noxalpha.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.\n\nOx 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.\n\nThe 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.\n\noxalpha.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.\n\n## Try it before the internet figures out what it is.\n\nFree while Ox Alpha stays in stealth. No account, no card, no waitlist.\n\n[Open chat](/chat)", "url": "https://wpnews.pro/news/ox-alpha-a-mysterious-new-ai-model", "canonical_source": "https://oxalpha.com", "published_at": "2026-08-25 09:31:53+00:00", "updated_at": "2026-08-25 09:45:11.154779+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models", "ai-products", "ai-research"], "entities": ["Ox Alpha", "fable-5", "glm-5.3", "gpt-5.6-sol", "grok-4.6"], "alternates": {"html": "https://wpnews.pro/news/ox-alpha-a-mysterious-new-ai-model", "markdown": "https://wpnews.pro/news/ox-alpha-a-mysterious-new-ai-model.md", "text": "https://wpnews.pro/news/ox-alpha-a-mysterious-new-ai-model.txt", "jsonld": "https://wpnews.pro/news/ox-alpha-a-mysterious-new-ai-model.jsonld"}}