Ox Alpha AI
Ox Alpha AI is a powerful reasoning model by Stealth — built for coding, agentic work, and complex tasks. Think step by step with a 1M token context.
Technical Specifications
Live data fetched from the OpenRouter model registry — always up to date.
What is Ox Alpha AI?
Ox Alpha is a stealth reasoning model operated by an anonymous third-party provider and routed through OpenRouter during its preview. It thinks through problems step by step before answering, making it exceptionally strong at logic, math, coding, and multi-step tasks. With a 1 million token context window, Ox Alpha can process entire codebases, long documents, and complex multi-file projects in a single session.
Features & Capabilities
Advanced Reasoning
Step-by-step logical thinking for complex problems — math, logic, and multi-step tasks.
Expert Coding
Write, review, debug, and refactor code across dozens of languages with ease.
1M Token Context
Handle entire codebases, long documents, and big data in a single session.
Agentic Workflows
Sustained multi-step tasks without losing track — built for production workloads.
Text + Vision
Combine written input with visual context for image understanding and analysis.
100% Free
No cost during the preview — chat unlimited on Hoodgen with your free credits.
Who Is Ox Alpha For?
Software Development
Architect solutions, debug stubborn issues, and review pull requests with deep code understanding.
Data & Analysis
Reason through datasets, build models, and explain complex analytical findings clearly.
Research & Learning
Get step-by-step explanations for hard concepts — like a personal tutor that never tires.
Production Workloads
Run long-horizon agentic tasks reliably — planning, execution, and verification in one flow.
Ox Alpha Pricing
- Unlimited chats on free plan credits
- No credit card required
- Included in Hoodgen AI free tier (100 credits/mo)
- Premium models also available on paid plans
What Makes Ox Alpha Different?
Ox Alpha launched anonymously on OpenRouter and OpenCode on August 20, 2026, under the provider name "stealth" with the model ID stealth/ox-alpha. No lab has officially claimed it, though community fingerprinting points strongly toward Z.ai's GLM family — a theory based on tokenizer probes that matched GLM 4 out of 4 times.
Three features define Ox Alpha's pitch. The first is the million-token context window — enough to hold entire repositories, documentation, and long task histories in a single session. The second is multimodal input covering text, image, and video, which is unusual for a coding-focused model. The third is the stealth preview itself: an anonymous launch with free usage during the preview window.
Ox Alpha also supports advanced agent features including tool calling, tool choice, structured output, reasoning effort control (max, high, low), and response format — making it a strong fit for production agentic workloads.
Ox Alpha Benchmark Performance
Early community benchmarks show promising results for coding tasks. In a viral 10-task DeepSWE test, Ox Alpha scored 80% — ahead of Claude Fable 5 at 65%, GLM-5.3 at 62%, and GPT-5.6-sol at 52%. However, a separate private benchmark rated it closer to older-generation models, so results remain contested until an audited leaderboard exists.
What Can You Do With Ox Alpha?
Testers have pushed Ox Alpha hardest on coding and generation. One user reported generating a GPU-accelerated fluid simulation as a single 1,000-line HTML file. Another asked for a dreamcore 3D world in three.js and received 64,745 tokens of procedural code from one prompt, with no external assets. Vision-grounded work is a recurring theme — developers have fed the model product videos and asked it to replicate apps, producing functional results.
Best Use Cases
- Long-repository exploration and architecture mapping
- Multi-file implementation with requirements alignment
- Visual bug triage — combining screenshots with code and error reports
- Low-cost evaluation of coding agents during the free preview
- Complex reasoning, math, and step-by-step problem solving
Ox Alpha Privacy & Data Policy
Ox Alpha is developed by an anonymous provider. On OpenRouter, prompts and completions are retained by the provider but not used for training. However, the broader OpenRouter Stealth Program agreement grants the provider broader collection and training rights, and stealth models can disappear at any time.
Because the provider is anonymous and access is temporary, we recommend treating Ox Alpha as an evaluation tool rather than a production dependency. Avoid pasting passwords, personal data, or proprietary source code during the free window. Hoodgen AI always keeps free alternative models available so you can switch without interruption.
Ox Alpha vs Other Models
Frequently Asked Questions
Ready to think deeper?
Chat with Ox Alpha free — sign in and start a conversation in seconds.
Start Chatting Free