This route is not a live model shootout. OpenAI’s Sora discontinuation notice says the web and app experiences ended April 26, 2026 and scheduled the API’s discontinuation for September 24, 2026; that date has passed as of this September 25, 2026 review. Google DeepMind currently presents Veo 3.1 as a video-generation model with audio and links to Gemini, Flow, and Veo development materials. Google’s model documentation also identifies Veo 3.1 and describes current video-generation workflows. This supports a status and workflow comparison only: we did not test either system, and do not claim that Sora remains available, that one makes better-looking video, or that either product grants commercial rights.
For a new generation workflow, evaluate the currently documented Veo 3.1 paths and verify which are enabled for your account; do not treat Sora as an active alternative after OpenAI’s announced shutdown dates. For footage already made in Sora, follow OpenAI’s current export instructions instead of assuming archived content will remain hosted. Google describes Veo 3.1 features including native audio and creative controls, but vendor descriptions are not a comparative quality result. Compare Veo with a currently available alternative on your own brief if output quality is the decision.
A product-status and documented-workflow update—not a model-quality benchmark
Sora: OpenAI Sora (openai-sora).
Veo: Google Veo (google-veo).
Availability for a new generation: Veo. OpenAI’s notice records the Sora web/app end date and scheduled September 24, 2026 API shutdown, which has passed at this review. Google currently documents Veo 3.1 and video-generation workflows. This is a product-status distinction, not proof that every Veo feature is available in every region or account; verify the live interface and entitlement before depending on it.
Documented video and audio workflow: Veo. Google describes Veo 3.1 as video generation with audio and documents several video workflows, including text-to-video, image inputs, reference videos, and video extension. The reviewed OpenAI notice is a discontinuation notice, not documentation for a currently usable Sora generation flow. These sources establish documented workflow scope, not comparative prompt adherence, audio fidelity, or visual quality.
Recovering existing Sora creations: Sora. For a prior Sora project, OpenAI’s discontinuation notice describes an export route and warns that associated data may be deleted after the applicable export window. That makes Sora relevant as an archive to retrieve, not as a current generator. Export timing is time-sensitive; use OpenAI’s live instructions and do not infer from archive access that generation is still available.
Evidence for a visual-quality winner: Even. The cited sources do not provide a controlled Sora-versus-Veo 3.1 comparison, and this guide did not run hands-on tests. OpenAI’s source establishes Sora’s discontinuation; Google’s pages describe its own model and workflows. Neither supports a current claim that one produces better-looking output. Test a currently available model against your own shot list and review the results.
Old Sora rankings describe a different availability context; OpenAI’s discontinuation notice is the source to consult for present status.
Google documents Veo 3.1 features and several video-generation paths, but access and eligible model versions can depend on the product surface and account.
OpenAI describes an export path for previous Sora creations and warns that associated data may be deleted after the applicable window; treat retrieval as time-sensitive.
This page makes no hands-on quality, price, or commercial-rights comparison. For production use, verify current provider terms and test outputs against your own brief.
Start a new video-generation project: Veo. Evaluate the Veo 3.1 path currently enabled for your account, confirm its input and output workflow in Google’s documentation, and test one representative shot before committing.
Retrieve an existing Sora clip or project: Sora. Follow OpenAI’s official Sora export instructions promptly and confirm the live export window. An archive/export route is not evidence that new Sora generation is available.
Generate video with audio in a currently documented workflow: Veo. Use Google’s Veo 3.1 documentation to identify the available product or development path, then confirm your account’s model access and review the result.
Decide whether a current model can replace a saved Sora shot: Either. Export the Sora shot if it is still available, make a Veo candidate from the same brief, and compare both against your actual edit requirements. This is a project-specific review, not a vendor-level quality verdict.
Do not select Sora for a new generation pipeline on the basis of older feature comparisons. OpenAI’s official notice says its web and app experience ended April 26, 2026 and gives September 24, 2026 as the scheduled API shutdown date. If you need older Sora work, treat this as a time-sensitive archive/export task and follow OpenAI’s current notice.
Shortlist Veo 3.1 for new work when its documented video-with-audio workflow fits. Google DeepMind describes native audio, extended videos, and creative controls, while Google’s model documentation identifies Veo 3.1 and video-generation workflows such as text, image, reference, and extension paths. Check the exact model, interface, account eligibility, and terms that apply before production; these documents do not establish a head-to-head quality winner.
OpenAI’s discontinuation notice says the Sora web and app experiences ended April 26, 2026 and scheduled the API’s discontinuation for September 24, 2026. That API date has passed as of this September 25, 2026 review, so this comparison does not present Sora as an active generation option. Check OpenAI’s current notice for any updated status before relying on access.
OpenAI’s notice describes an export route for existing creations and warns that associated data may be deleted after the applicable export window. Follow the live instructions at OpenAI rather than assuming old content remains hosted; retrieving an archive is separate from using Sora to generate new video.
Google DeepMind presents Veo 3.1 as video generation with audio and describes native audio, extended videos, and creative controls. Google’s model documentation identifies Veo 3.1 and provides video workflows. These pages describe product capabilities, not a guarantee of access for every account or a comparative quality score.
The reviewed evidence cannot answer that as a current head-to-head question: OpenAI’s notice documents Sora’s discontinuation, and we did not run hands-on tests. Google’s own Veo material describes its product but is not an independent comparison with an available Sora model. For a live project, compare Veo with another currently available option using your own prompts and review criteria.