Robots Atlas>ROBOTS ATLAS
Ox Alpha
AI Models

Ox Alpha

alpha (preview)
Anonymous "stealth" reasoning model offered free on OpenRouter as stealth/ox-alpha; for coding, long-horizon agentic work and production. Creator undisclosed.
⏳ Preview⏳ Limited accessReasoning modelMultimodalTool-using model
Context window
1M
tokens
Max output
131,072
tokens
Release date
20 August 2026
Access:APIHostedDeployment:☁ Cloud

Overview

Ox Alpha is a "stealth" AI model that appeared on OpenRouter on 20 August 2026 under the identifier stealth/ox-alpha. Its creator has not disclosed their identity — the model is described as "developed and operated by a third-party provider who has chosen to remain anonymous".

It is positioned as a reasoning model built for coding, sustained agentic work and production workloads, particularly long-horizon software engineering that combines text with visual context.

The model offers a 1,048,576-token (1M) context window and up to 131,072 output tokens. It accepts text, image and video input (audio requests are rejected), and supports function calling (tool use) and JSON-formatted responses.

During the preview phase it is available for free via the OpenRouter API and in the OpenCode tool. Under OpenRouter's terms, prompts and completions are retained though not used for training — raising privacy concerns when handling sensitive code.

The creator's identity remains officially unconfirmed. Speculation links the model to the Chinese company Z.ai (formerly Zhipu AI) and its GLM family, or to an unreleased Microsoft MAI model, but there is no public evidence identifying the responsible organization.

Classification
Reasoning modelMultimodalTool-using model
Access & deployment
APIHosted
Cloud
Weights: Closed
Key parameters
📏 Context: 1M
✓ Tools
📥 Input: text, image, video

Technical specification

Context window
1M
tokens
Max output tokens
131,072
tokens per response
Features:✓ Tool use
Modalities
⬇ Input
textimagevideo
⬆ Output
textcode

Capabilities and applications

Native model capabilities
Coding
Generating, analysing and modifying code in many programming languages. Covers writing functions, debugging, refactoring, code review, and creating tests. Measured by benchmarks such as HumanEval and SWE-bench.
Category: coding
Agentic coding
Multi-hour, multi-step programming tasks performed autonomously by the model: cloning a repository, running tests, iterating on fixes, integrating with CLI tools. Characteristic of Codex variants (GPT-5.1-Codex-Mini, Codex-Max).
Category: coding
Agentic capability
The model's ability to autonomously plan and execute multi-step tasks by sequentially using tools, maintaining context, and adapting to intermediate results.
Category: planning
Long context
Support for large context windows — tens to hundreds of thousands (or millions) of input tokens. Enables analysis of entire codebases, long documents, and many parallel conversations without losing earlier information. GPT-5.1 supports 400,000 tokens.
Category: language
Advanced reasoning
The ability to perform multi-step, structured reasoning: analysing problems, planning steps, and drawing conclusions from hypotheses. Reasoning-first models (e.g. GPT-5.1 Thinking) dedicate a portion of inference to chains of thought before responding.
Category: reasoning
Image understanding
Analysing and interpreting the content of images.
Category: vision
Function Calling
Category: planning

Pricing

Technical architecture

Sources and related pages