Redditor Tanmay_Vermaa's dataset check found 31 of 70 third-party apps still offering Sora after its September 24 API shutdown, implying retired-model access through unofficial or misrepresented channels; verifying what those apps actually serve would establish post-sunset model-identity deception as a live consumer-trust issue.
state: seedheat: lowuncertainty: mediumconvergesscott: highmodel-identity api-sunset inference-providers
What is this?
OpenAI wound down Sora in two stages: the consumer app and sora.chatgpt.com closed April 26, 2026, and the developer Videos API (sora-2, sora-2-pro) shut September 24, 2026, with no successor model announced. The anchor evidence is a Reddit dataset check by Tanmay_Vermaa whose title reports that after that date, 31 of 70 third-party apps listing Sora still offered it, implying unofficial or misrepresented access to a retired model. The supplied web results corroborate the shutdown timeline and confirm a third-party resale layer existed around Sora — one platform still markets a 'Sora 2 generator' as a stable third-party entry, and some aggregators paused their Sora endpoints ahead of the deadline — but none of the snippets independently verify the 31/70 measurement or establish what those apps actually serve post-sunset; that verification is precisely the open step the case proposes.
Why it matters to Scott
Because the Sora API is provably dead, every one of the 31/70 apps still 'offering' it makes a falsifiable model-identity claim — exactly the 'independently checkable, not asserted' thesis of his cryptographic-trust and version-bound-assessment work, and the proposed verification (fingerprint what they actually serve) is the same probe-providers-and-check harness genre as his remote-exec benchmark. It extends the silent-substitution lineage the radar already tracks (CrofAI reseller substitution, OpenAI Astra rerouting) into the consumer wrapper layer with a quantified, publishable receipt for his AI-washing and model-perishability arguments.
ip:concept.cryptographic-trustdev:concept.version-bound-ai-assessmentdev:project.remote-execip:concept.model-perishabilitydev:concept.owned-output-model-levelsip:source.why-ai-projects-are-failing-explained-ebookradar:concept.model-fingerprintingradar:concept.model-provenanceradar:one-token-api-model-fingerprintingradar:crofai-model-substitution-shutdownradar:openai-astra-silent-rerouting-claim
queries asked of Scott's wikis
- model identity verification fingerprinting what model served output
- silent model swap provider substitution misrepresentation
- API deprecation sunset agent tool dependency breakage handling
- AI wrapper app misleading marketing dark patterns
- inference aggregator proxy reseller routing trust
- model provenance audit harness output fingerprint
Measured heat
now 0 pts/hpeak 0 pts/hcomments 0/hpeers p0momentum: steady1 platformsage 314h
points/hour across evidence · reading as of 2026-10-12 02:59:37.977291+11:00 · deterministic, not a model opinion
How the heat travelled
pace: p22 vs 1188 stories at the 168h mark (now 314h old) — ahead of aafp-commons-signed-agent-notebook (2.0x), behind agentgate-signed-agent-receipts (0.7x)
Evidence (1) — ⭐ canonical anchor
Interpretation history
2026-09-28T16:07:08Z
grounded: converges/high — Because the Sora API is provably dead, every one of the 31/70 apps still 'offering' it makes a falsifiable model-identity claim — exactly the 'independently che
2026-09-28T15:57:22Z
case created — Concrete, checkable measurement of a distinct app-layer post-sunset model-identity episode, separate from the open provider-side rerouting and substitution cases.
Decision trace
- 09-29 02:07groundBecause the Sora API is provably dead, every one of the 31/70 apps still 'offering' it makes a falsifiable model-identity claim — exactly the 'independently checkable, not asserted'
- 09-29 01:57createConcrete, checkable measurement of a distinct app-layer post-sunset model-identity episode, separate from the open provider-side rerouting and substitution cases.