Sora Alternatives After Shutdown: Free, Mobile, Local and API Picks
Sora's app, web experience and API are discontinued. This guide shows how to export existing work, separates Runway's limited no-payment trial from paid Gen-4.5 access, and gives picks for mobile creation, multi-shot dialogue, local generation and API workflows. Prices checked September 26, 2026.
Contents

Sora’s app and web experience are gone, and its API has also reached its shutdown date. Your first move is not to compare models: it is to export anything you still have in Sora. After that, choose a replacement by the job you actually need to finish—free testing, phone-based creation, multi-shot storytelling, local generation or an API pipeline.
The direct picks are: Runway Free only as a limited no-payment workflow sampler—the included models can change and Gen-4.5 requires Standard or higher—Runway for mobile creation, Kling VIDEO 3.0 for multi-shot scenes and dialogue, Google’s Gemini API for Veo-specific controls or conversational video editing, and LTX Desktop for supported local hardware. Prices and access below were checked on September 26, 2026.
Export your Sora work before choosing a replacement
OpenAI says the Sora web and app experiences were discontinued on April 26, 2026, and the Sora API was discontinued on September 24, 2026. The same official discontinuation notice says data will be permanently deleted after discontinuation and any final export window that OpenAI may be able to offer.
Do this now:
- Open sora.chatgpt.com/sunset.
- Select Export.
- Watch the email address on your OpenAI account for the download notification.
- Download the archive, verify that it opens, and keep a second copy in another storage location.
OpenAI does not publish a guaranteed closing date for that possible final export window. It says users will be notified by email if one is offered and recommends exporting as soon as possible. Do not wait for a countdown page.
The best Sora replacement depends on what you used Sora for
| Your main job | Start with | Why this is the first test | Current access and cost boundary |
|---|---|---|---|
| Sample the workflow without paying | Runway Free (limited trial) | Its one-time credits are documented, but the included model set can change | 125 credits once; check the logged-in model picker; Gen-4.5 requires Standard or higher |
| Create from an iPhone or Android phone | Runway | Official apps exist for both platforms | Mobile works, but Runway recommends desktop or laptop for the full experience |
| Generate a short story with several shots and spoken dialogue | Kling VIDEO 3.0 | Multi-shot planning, native audio, character references and up to 15 seconds are built into the current model | 6–12 credits per second before optional voice control, depending on resolution and audio |
| Edit video conversationally with several kinds of reference input | Gemini Omni Flash | Google recommends it as the default video model for coherence, multi-input reasoning and multi-turn edits | About $0.10 per second of 720p API output; no video free tier in the Gemini API |
| Extend a scene or control the last frame | Veo 3.1 | Google specifically positions Veo for scene extension, last-frame control and legacy pipelines | Lite starts at $0.05 per second; Fast at $0.10; Standard at $0.40 for 720p |
| Run an open-source desktop workflow locally | LTX Desktop | It can generate locally instead of sending each generation to a cloud service | Requires supported hardware; the app is beta and model weights may have separate terms |
| Use one developer API for several current video models | Runway Dev | One billing system exposes Runway and selected third-party models | API credits cost $0.01 each; model rates differ |
These are scenario picks, not a universal visual-quality ranking. There is no shared first-party benchmark here that runs the same prompt, reference files and settings across every service.
Use Runway Free to sample the workflow, not to test Gen-4.5
Runway Free is useful for learning the interface and generation flow, but it is not a free Gen-4.5 entry point. Runway’s Free plan details say the plan provides 125 one-time, non-expiring credits and access to a limited model set that changes over time; log in to see the current list. Free outputs carry a watermark, and Free accounts cannot buy extra credits.
Runway’s Gen-4.5 specification requires Standard or higher and charges 12 credits per generated second: 60 credits for 5 seconds and 120 for 10. That calculation belongs to paid plans, so the Free plan’s 125-credit balance must not be presented as one 10-second or two 5-second Gen-4.5 clips.
Runway’s individual plans currently list Standard at $15 monthly or an effective $12 per month with annual billing, Pro at $35 or $28, and Max at $95 or $76. Standard includes 625 monthly credits, Pro 2,250 and Max 9,500. Standard and Pro credits reset rather than roll over; Max can roll over up to one month.
Use Free only to check the models your account currently exposes and to learn the prompt, generation and export flow. An upgrade prompt means that model is not in Free; to evaluate Gen-4.5, budget 60 or 120 paid-plan credits on Standard or higher.
The no-payment boundary for Runway, Google and Kling
Google’s Gemini API pricing marks the video free tier as unavailable for Gemini Omni Flash and Veo 3.1. Google’s consumer plans do include video generation in Gemini and Flow, but that is a subscription route rather than a guaranteed free API allowance.
Kling’s developer guide mentions a trial resource package for integration testing, but the public page reviewed does not state a durable quantity or a recurring consumer grant. Treat any credits shown in your Kling account as an account-specific offer and confirm them before planning a project.
LTX Desktop can avoid a cloud generation charge when it runs locally, but “local” is not the same as “free.” You still need capable hardware, storage, electricity, model downloads and setup time.
For mobile creation, choose Runway first
Runway’s mobile compatibility page explicitly lists official iPhone and Android apps. That makes it the lowest-friction replacement when Sora was mainly a phone workflow for quick prompts, image-to-video experiments and reviewing drafts away from a desk.
Use the phone app to start and review generations, but move to a desktop or laptop when you need the full interface, larger asset management or a more involved production workflow. Runway itself recommends desktop or laptop for the complete experience, and its iOS app does not support iPad.
Google is the second mobile route when you already pay for a Google AI plan. Google’s US plan page includes video generation in Gemini and Google Flow; it lists Google AI Pro at $19.99 per month and Ultra at $99.99 per month. Capacity and feature eligibility can vary, so confirm the current limits in your account rather than assuming a fixed number of clips.
Kling’s VIDEO 3.0 guide documents an app-only character-recording workflow, so an app experience exists. However, the public guide reviewed does not give a stable global app-store availability matrix or a recurring free allowance. For a mobile-first decision, Runway has the clearer official contract.
For multi-shot scenes and dialogue, choose Kling VIDEO 3.0
Kling VIDEO 3.0 is the first model to test when a single Sora prompt used to represent a miniature scene rather than one continuous visual moment. Its official model guide documents automatic and custom multi-shot generation, native audio, start-and-end-frame input, character references and clips from 3 to 15 seconds.
The same guide supports dialogue in Chinese, English, Japanese, Korean and Spanish, including several speakers and specified accents. That makes Kling especially relevant for a short ad, dialogue exchange, storyboard preview or character scene where cuts and speech need to be planned together.
Current VIDEO 3.0 credit rates are:
| Kling VIDEO 3.0 mode | 720p | 1080p | Ten-second example |
|---|---|---|---|
| No native audio | 6 credits/s | 8 credits/s | 60 or 80 credits |
| Native audio | 9 credits/s | 12 credits/s | 90 or 120 credits |
| Voice control add-on | +2 credits/s | +2 credits/s | +20 credits |
Do not convert these figures to dollars using a third-party estimate. The public model guide states credit consumption, while the cash price of credits can depend on the package and account shown in Kling’s current purchase interface.
Choose Kling when multi-shot narrative structure and dialogue are central. For simple image-to-video, check which model your Runway Free account currently exposes, but do not treat it as a free Gen-4.5 trial. Google offers clearer per-second API dollar pricing.
For local or open-source use, choose LTX Desktop—only with suitable hardware
LTX Desktop is the clearest replacement here when your priority is a local workflow, inspectable application code or keeping generation on your own machine. The official LTX Desktop repository describes an open-source desktop app for text-to-video, image-to-video and audio-to-video generation, with local models and an API fallback.
Local mode has a real hardware floor:
- Windows or Linux: an NVIDIA CUDA GPU with at least 16 GB of VRAM. Windows also calls for 160 GB or more of free disk space.
- Apple Silicon Mac: macOS 13 or newer and at least 15 GB of free RAM at launch, not merely 15 GB installed.
- Unsupported or underpowered hardware: the app falls back to API-only mode.
The desktop app is licensed under Apache-2.0, but downloaded model weights can have additional license terms. It is also marked Beta, so expect changes. LTX 2.5 Fast is the default local model; Pro variants are API-only, and some operations remain tied to older local checkpoints or cloud models.
Choose LTX Desktop if you already have the hardware, accept model downloads and want local control. If buying a GPU only for occasional short clips, a cloud plan or per-second API is usually the simpler first experiment.
For an API, choose between Google direct access and Runway Dev
Use the Gemini API directly when your workflow depends on Google’s current model-specific controls. Google’s video generation guide recommends Gemini Omni Flash as the default for multi-input reasoning, character consistency and conversational edits. It positions Veo 3.1 for scene extension, last-frame control and integration with existing Veo pipelines.
The current paid API rates are:
| Google model and output | Price per second | Ten-second output |
|---|---|---|
| Gemini Omni Flash, 720p effective rate | about $0.10 | about $1.00 |
| Veo 3.1 Lite, 720p | $0.05 | $0.50 |
| Veo 3.1 Lite, 1080p | $0.08 | $0.80 |
| Veo 3.1 Fast, 720p | $0.10 | $1.00 |
| Veo 3.1 Fast, 1080p | $0.12 | $1.20 |
| Veo 3.1 Standard, 720p or 1080p | $0.40 | $4.00 |
| Veo 3.1 Standard, 4K | $0.60 | $6.00 |
Google says a failed generation caused by the documented audio-processing issue is not charged. Video models are not available on the Gemini API free tier.
Choose Runway Dev when you would rather integrate one API and compare several models under one credit system. API credits cost $0.01 each. Gen-4 Turbo is 5 credits per second, or $0.05 per second; Gen-4.5 is 12 credits, or $0.12 per second. Runway’s current catalog also lists Veo 3.1, Gemini Omni Flash and other providers, with the response reporting the model selected and the realized credit cost when a router is used.
Choose the direct Google API when a Google-specific feature is a requirement. Choose Runway Dev when model switching and unified billing are more valuable than being first to every vendor-specific capability.
Kling also has an official developer platform and a global endpoint for servers outside China. It is the API to test when Kling’s multi-shot, character or dialogue behavior is the reason for migrating. Confirm the current resource-package price and quota inside the developer console; the public docs do not give a stable dollar-per-credit rate.
Use one controlled shot to compare replacements
Do not move an entire Sora workflow after watching unrelated showcase clips. Create a small comparison task that matches your real work:
- Pick one five- to ten-second shot you genuinely need.
- Reuse the same prompt, aspect ratio, duration and reference image wherever the tools allow it.
- Record the number of generations required to get an acceptable result—not only the advertised price of one generation.
- Check whether audio, text, character appearance and the final frame survive the workflow you need.
- For APIs, record task creation, polling, failure handling, output retention and the actual billed cost.
- Read the current commercial-use and model-weight terms before using output for client work.
A cheaper per-second rate can cost more if it takes many retries. A subscription can be wasteful if you only need a few clips. A local model can reduce cloud charges but increase setup and rendering time. Use the smallest realistic shot to expose those trade-offs before committing.
Final picks after Sora’s shutdown
- Export now: use OpenAI’s Sora sunset page; there is no guaranteed published final-window deadline.
- No-payment workflow sample: Runway Free has 125 one-time credits, but the included models can change; check your account, and use Standard or higher for Gen-4.5.
- Best mobile starting point: Runway, because both iPhone and Android apps are officially documented.
- Best for multi-shot dialogue: Kling VIDEO 3.0, because shot planning, native audio and multilingual dialogue are part of the current model contract.
- Best for conversational video editing: Gemini Omni Flash through Google’s consumer tools or paid API.
- Best for Veo-specific control: Veo 3.1 through the Gemini API.
- Best local option: LTX Desktop on supported NVIDIA or Apple Silicon hardware.
- Best multi-model developer route: Runway Dev; use a direct vendor API when a vendor-specific feature is non-negotiable.
Start with the category that matches your former Sora workflow, not the model with the longest feature list. The right replacement is the one that completes your actual shot with an acceptable number of retries, at a cost and access method you can sustain.