No — the free preview ended on August 26, 2026. Ox Alpha was revealed as Z.ai’s GLM-5.3-Flash and is now a paid model. There is promotional pricing until September 9, 2026, and there are still genuinely free ways to use a capable model, both covered below. If you had a project running on stealth/ox-alpha, the migration path is at the bottom of this page.
Status right now
Live: stealth models on OpenRouter right now
| Model | Listed since | Context | Price (in / out) | Input | Status |
|---|---|---|---|---|---|
| No stealth models are listed at the moment. | |||||
Stealth reveal history
| Stealth codename(s) | Revealed as | When |
|---|---|---|
| Quasar Alpha, Optimus Alpha | GPT-4.1 | April 2025 |
| Sonoma Sky Alpha, Sonoma Dusk Alpha | Grok 4 Fast | September 2025 |
| Pony Alpha | GLM-5 | February 2026 |
| Hunter Alpha, Healer Alpha | Xiaomi MiMo-V2 | March 2026 |
| Elephant Alpha | Ling-2.6-flash | April 2026 |
| Owl Alpha | LongCat-2.0 | June 2026 |
| Ox Alpha | GLM-5.3-Flash (Z.ai) | August 2026 — listed Aug 20, revealed Aug 26 |
What it costs now
| Tokens | Promotional (until Sept 9, 2026) | List price (after) |
|---|---|---|
| Input, per 1M | $0.075 | $0.15 |
| Cached input, per 1M | $0.015 | $0.03 |
| Output, per 1M | $0.25 | $0.50 |
The promotion ends at 24:00 UTC+8 on September 9, 2026, at which point prices double. That is still inexpensive by any reasonable comparison — worked examples against typical daily workloads are on the pricing page.
Timeline
| Date (UTC) | Event | Status |
|---|---|---|
| Aug 20, 2026 | stealth/ox-alpha listed on OpenRouter under the anonymous “Stealth” provider: $0 pricing, 1,048,576-token context, text/image/video input. | Confirmed |
| Aug 20–21 | Developers push the model through coding agents; fingerprinting threads begin; press reports a hosting partner describing a one-week free window. | Reported |
| Aug 21 | Ben Davis posts a 10-task DeepSWE subset result (~80%). | Third-party |
| Aug 22–23 | Full 113-task DeepSWE run reported at ~63%. Details. | Third-party (secondary reports) |
| Aug 20–26 | Ox Alpha becomes the most-used model on OpenRouter during its preview. | Reported |
| Aug 26, 2026 | Revealed as Z.ai GLM-5.3-Flash. Bloomberg reports the identity (~09:00 UTC), the OpenRouter catalogue updates to z-ai/glm-5.3-flash (13:59 UTC), Z.ai announces that evening. Free preview ends; weights published under MIT. | Confirmed |
| Sept 9, 2026 | Promotional pricing ends at 24:00 UTC+8; list pricing takes effect. | Announced |
The preview ran six days, close to the “one week” a hosting partner had described to the press on day one. Our standing expectation on this page was that it would end around August 27 — it ended a day earlier, alongside the reveal rather than before it.
If you were using stealth/ox-alpha
The old model ID is delisted and no redirect exists — calls to stealth/ox-alpha now fail rather than silently rerouting. Replace it with z-ai/glm-5.3-flash on OpenRouter, or glm-5.3-flash if you are calling the Z.ai API directly. Because it is now a paid model, you also need credit on the account behind your key; a $0 balance that worked fine during the preview will start returning errors. The API and migration guide covers this in detail.
Free ways to use it, and free alternatives
- Run it yourself. The weights are on Hugging Face as
zai-org/GLM-5.3-Flashunder the MIT licence. Free in licensing terms; you still pay for hardware. Deployment paths include SGLang, vLLM, TokenSpeed, KTransformers and others. - Use a free model instead. OpenRouter maintains a rotating set of genuinely free models, and the chat on this site lets you pick one from the model switcher — no credit required.
- Z.ai’s own free tier, where offered. Check docs.z.ai for current terms; these change often enough that we do not restate them here.
Our setup guide walks through both the paid and the free routes step by step.
What past stealth models did
Seven OpenRouter stealth codenames have now been revealed under official names, from several different developers; in each case the codename listing was replaced by the official, priced model. The full list is in the reveal history table. What has never happened: a stealth model simply vanishing without identification.
FAQ
Is Ox Alpha still free?
No. The free preview ended on August 26, 2026, when the model was revealed as GLM-5.3-Flash. It is now paid, at $0.075 per million input tokens and $0.25 per million output tokens until September 9, 2026.
How long was Ox Alpha free?
Six days — from August 20 to August 26, 2026.
Can I still use GLM-5.3-Flash for free?
Not through OpenRouter’s paid listing. You can run the MIT-licensed weights on your own hardware, or switch to one of OpenRouter’s rotating free models, which the chat on this site supports.
Does stealth/ox-alpha still work?
No. The ID is delisted and there is no redirect. Use z-ai/glm-5.3-flash instead.
Sources and verification
Last substantively reviewed: August 27, 2026. Reviewed by: OxAlpha.chat Editorial Team.
How we verified this page
The status of the former free preview was checked against the reveal date, current platform listing and official pricing. “Free weights” means no license fee; it does not mean that hosting hardware or third-party inference is free.
Primary sources
- Official reveal: Z.ai GLM-5.3-Flash announcement — confirms the release that ended the anonymous preview.
- Official pricing: Z.ai API pricing — provides the paid promotional and standard rates.
- Live platform record: OpenRouter GLM-5.3-Flash listing — shows the current priced model and providers.
- Platform billing guidance: OpenRouter FAQ — explains credits, free-model limits and usage billing.
- License: Official MIT license file — supports the distinction between open weights and free hosted inference.
Found a factual error or a changed price? Send us the page URL and supporting source; corrections follow our editorial policy.