Ox Alpha appeared as a stealth model and quickly drew attention for coding and agent work. The official OpenRouter catalog confirms a model ID, supported inputs, tool capabilities, context size, output limit, and current price. It does not identify the maker. Search results and social posts often mix those confirmed fields with unsupported guesses about the lab behind it.
Use this guide if you are evaluating Ox Alpha for coding, repository analysis, image or video understanding, tool calling, or long-running agent tasks and need a factual starting point before testing it.
Quick answer
Ox Alpha is the OpenRouter model identified as stealth/ox-alpha. At the time checked, the catalog describes it as a reasoning model for coding and sustained agentic work, accepts text, image, and video input, supports tool calls and structured output, provides a 1,048,576-token context window, and lists a 131,072-token maximum completion. Access and pricing can change. The maker is not confirmed in the official listing, so claims that name a specific company should be treated as theories unless the provider discloses itself.
What to check first
- Open the live OpenRouter catalog and locate the exact stealth/ox-alpha model ID.
- Record context length, maximum completion, input modalities, supported parameters, price, and moderation status with a UTC timestamp.
- Check whether the route is available before designing a workflow around it.
- Run one short coding task, one repository task, one tool call, and one structured-output task with fixed fixtures.
- Keep confidential repositories and client data out until provider identity, retention, and contractual terms are acceptable.
Diagnostic table
Use this table to connect the observed behavior to evidence and a verification step.
| Action | Evidence to collect | How to verify |
|---|
| Freeze the official model metadata | Open the live OpenRouter catalog and locate the exact stealth/ox-alpha model ID. | Every factual capability maps to the current official catalog. |
| Separate facts from theories | Record context length, maximum completion, input modalities, supported parameters, price, and moderation status with a UTC timestamp. | Unknown ownership is stated without implying a favorite theory is confirmed. |
| Run fixed representative tasks | Check whether the route is available before designing a workflow around it. | Results include prompt, harness, settings, attempts, cost, and timestamp. |
| Review privacy and routing | Run one short coding task, one repository task, one tool call, and one structured-output task with fixed fixtures. | The production workflow has time, token, privacy, and fallback limits. |
Why this usually happens
- Stealth previews are intentionally listed before a provider reveal.
- Free access and a large context window create rapid experimentation and rumor.
- Different evaluations use different harnesses, prompts, tools, budgets, and success criteria.
- A model can be strong on one repository task and still fail simple deterministic checks.
Useful command or data shape
Adapt paths, IDs, and privacy handling to the site before running commands or storing data on production.
model_id: stealth/ox-alpha
checked_at_utc: 2026-08-24T13:00:00Z
input_modalities: [text, image, video]
output_modalities: [text]
context_tokens: 1048576
max_completion_tokens: 131072
maker: not disclosed in official catalog
Test scenarios to run
Run the same controlled fixture across these branches. Write down the expected result before testing so a surprising response is easy to identify.
| Scenario | Fixture | Expected result |
|---|
| Catalog identity | stealth/ox-alpha | Exact model ID is present |
| Long context | Known files plus distractors | Cites the right file and line |
| Tool call | Read-only schema | Valid arguments, no invented tool |
| Structured output | Strict JSON schema | Parses without repair |
Safe fix order
Use a sequence that makes each result easy to prove. Stop when new evidence changes the scope or owner of the problem.
- Freeze the official model metadata
- Separate facts from theories
- Run fixed representative tasks
- Review privacy and routing
- Set a fallback and spending cap
Decision rule
Use Ox Alpha for a bounded trial only when the live route is available, your fixtures pass, privacy requirements permit the provider route, and a known fallback can take over without losing work.
Production verification checklist
- Every factual capability maps to the current official catalog.
- Unknown ownership is stated without implying a favorite theory is confirmed.
- Results include prompt, harness, settings, attempts, cost, and timestamp.
- The production workflow has time, token, privacy, and fallback limits.
Field notes
- Separate confirmed facts, reproducible observations, and community theories with explicit labels.
- Use a dated test fixture because model behavior, prices, leaderboards, bids, and traffic can change between visits.
- Keep the first answer concise, then show the evidence, failure cases, and decision rule that support it.
- Link to the closest next diagnostic instead of sending every reader to a generic service page.
Questions teams ask during testing
How often should this page be checked?
Review the official source whenever access, ownership, pricing, limits, auction rules, or live counters change. Keep dated observations separate from durable guidance so one update does not invalidate the whole article.
Can this information be used for a client decision?
Use it as a starting point, then verify the current official page and run a small controlled test. Do not send confidential code, personal data, credentials, or irreversible spend into a new service based only on a public trend.
What should a useful test record?
Keep the UTC time, exact URL or model ID, input or campaign parameters, expected result, actual result, cost, latency, downstream record, and the person who decides whether to continue.
How should this connect to measurement?
Give every campaign one tagged destination, preserve the parameters through the landing flow, store the values with the lead or order, and reconcile the browser record with the CRM or commerce record.
Mistakes to avoid
- Repeating a rumor as a confirmed fact when the model, domain, auction, or leaderboard can change faster than an article is indexed.
- Publishing a broad definition without a test, calculation, checklist, source, or decision that gives the reader a reason to keep the page open.
- Sending campaign traffic to an untagged URL, then trying to reconstruct source and creative details from incomplete analytics later.
- Copying live counters into evergreen copy without a UTC timestamp, source link, and clear warning that the number will change.
What to tell the client or owner
Record the current source URLs, UTC test time, exact fixture, expected and actual outcome, spend or token usage, privacy classification, owner, next review date, and the metric that would change the decision. Redact secrets, personal data, client code, and account identifiers.
When HandL WP should help
Bring in help when this affects leads, checkout, search visibility, security, paid media reporting, or a client production site. HandL WP can trace the issue through WordPress, hosting, cache, tracking, and Search Console, then verify the workflow after the technical fix.
If this is active on a production site, review an AI-assisted WordPress workflow.
Related HandL WP guides
Use these related guides when the same issue touches tracking, security, checkout, or crawler visibility.
Test capability, retrieval, and total task cost
Move beyond the catalog with the Ox Alpha tool-call reliability harness, the million-token retrieval test, and the free-preview total task cost audit.
Helpful references