Independent YFarmX measurements indicate that Ox Alpha is consistent with the Z.ai family at very high confidence. These measurements establish lineage evidence only: a shared tokenizer or serving contract shows shared ancestry, and does not independently establish the precise underlying release.
| Signal | Result | Reference |
|---|---|---|
| Tokenizer signature | Exact match with 2 other models | tk_133b086c |
| API surface | Identical contract to 2 other entries | api_7e146681 |
| Context and output | 1,048,576 context · 131,072 max output | declared |
| Reasoning contract | Mandatory · efforts max, high, low · default max | declared |
| Serving providers observed | Stealth | measured 21 August 2026 |
Click a row to open the full comparison.
An x marks a row excluded as corrupt (caching or provider interference detected during measurement). Overhead subtracted: 87 prompt tokens.
Ox Alpha appeared on OpenRouter on 20 August 2026 as a stealth entry: no lab, no family, a cloaking system prompt, and a catalogue tag of Other. Two independent measurements were taken while it was live. Across the 418 models then listed, exactly three carried its complete API contract: Ox Alpha itself, GLM-5.3, and the GLM-latest alias. GLM-5.2 shares only two of the seven contract fields, so the contract appeared with the 5.3 generation rather than being a standing template.
The tokenizer measurement is the part a copied catalogue record cannot explain, because token counts come from the inference backend. Ox Alpha agreed with GLM-5.3 on all fifty adversarial strings. The load-bearing control is Kimi K3: it carries the same Other tag, so if OpenRouter fell back to a shared estimator for that tag, Kimi would also have scored near 100%. It scored 18%.
What this supports: shared lineage with the Z.ai GLM family, at very high confidence. What it does not support: that Ox Alpha is GLM-5.3. Ox accepts image and video input while public GLM-5.3 is text-only, and a shared tokenizer is family evidence, not a release identification. If an official identity is announced, this record keeps the prediction as it stood.
Ox Alpha is a reasoning model designed for coding, sustained agentic work, and production workloads. It is suited for long-horizon software engineering, complex reasoning, and workflows that combine text with...
Supported parameters
include_reasoningmax_tokensreasoningreasoning_effortresponse_formattemperaturetool_choicetoolstop_ktop_p
defaults: {"temperature":1,"top_p":0.95}
Cite this page as the evidence record for Ox Alpha: the URL is stable, measurements are dated, and revisions append to the history above. Method: tokenizer fingerprinting, v1, 50 strings.
