Runway Gen-4: the complete guide
Runway Gen-4 is Runway's general-purpose video generation family inside a full creative suite — not just an API endpoint. The current flagship is Gen-4.5 for text-to-video and image-to-video; Gen-4 Turbo is the budget image-to-video tier. This guide pulls specs and credit math from Runway Academy and the Developer API pricing page (July 2026). We feature Runway in several property and Instagram ad battles on PromptBattleLab; battle verdicts with logged costs are on /models/battles — this page is the documentation and billing reference for indexing.
Quick facts
- Developer:
- Runway (runwayml.com)
- Current video models:
- Gen-4.5 · Gen-4 Video · Gen-4 Turbo · Gen-3 Alpha Turbo (legacy)
- Modes:
- Text-to-video, image-to-video, video-to-video (Aleph), character (Act-Two)
- Duration:
- Up to 10 s per gen on paid plans (varies by model)
- Resolution:
- 720p–1080p export; 4K upscale on higher plans
- Native audio:
- Platform tools (voice, stitch, extract) — not native in all video gens
- API access:
- Runway Developer API (separate credit wallet from web app)
- API price from:
- 5 credits/s (Gen-4 Turbo) = $0.05/s · 12 credits/s (Gen-4.5) = $0.12/s
What Runway Gen-4 actually is
Runway positions Gen-4 as a general-purpose video generator inside an editing product: motion brush, camera controls, inpainting, upscale and stitch flows sit beside raw generation. That matters for agencies — you may pay more per second than fal-only pipelines, but you also skip round-tripping through After Effects for simple fixes.
- ▸Gen-4.5 (latest): highest realism and cinematography in Runway's lineup — 12 credits per second on API.
- ▸Gen-4 Video / Gen-4 Turbo: image-to-video focused; Turbo at 5 credits/s is the budget API path.
- ▸Gen-4 Aleph / Aleph 2.0: video-to-video editing and stylization — different pricing (15–28 credits/s).
- ▸Act-Two: character performance animation from a reference performance clip — 5 credits/s, 3 s minimum.
- ▸Web app subscriptions (Standard $28/mo, Pro $76/mo, etc.) bundle credits separate from Developer API credits.
Pricing — credits, subscriptions and API
Developer API: 1 credit = $0.01. Video cost = credits per second × duration × $0.01.
Web app subscriptions amortize differently — a Pro plan at $28/mo with 2,250 credits makes each credit ~$0.0124, so Gen-4 Turbo at 5 credits/s is ~$0.062/s effective before you count unused credits. API pay-as-you-go is simpler for automation: $0.05/s Turbo, $0.12/s Gen-4.5.
- ▸Free tier: 125 one-time credits — enough for short tests, not production battles.
- ▸Unlimited plan ($188/mo): unlimited Gen-3 generations plus a Gen-4 credit pool — read the fine print before assuming unlimited Gen-4.5.
- ▸Upscale to 4K: flat ~50 credits on web app — factor into total deliverable cost.
- ▸No public affiliate program as of July 2026.
API parameters and integration
Runway Developer API uses task-based async jobs. Model selection happens via the model field on the generation task — exact schema is in the Runway API docs. Typical inputs:
Authentication: API key from the Runway Developer Portal. Poll task status until SUCCEEDED; download output URL. Failed tasks may still consume credits depending on failure code — safety blocks on input are documented as non-refundable in third-party API guides; verify in your dashboard.
Prompting guide — platform strengths
- ▸Instagram and vertical ads: Runway's docs emphasize polished motion on fashion, lifestyle and product — start from a strong still for Gen-4 Turbo i2v.
- ▸Image-to-video from listing photos: crop to 9:16 with safe margins; Turbo at $0.25 per 5 s is a common agency workflow.
- ▸Use motion brush and camera controls in the web app when API output is 80% right — the suite value is fixing in-place.
- ▸Gen-4.5 text-to-video: write lighting and lens feel explicitly; Runway markets this tier for "high realism and cinematography".
- ▸Avoid architecture-only prompts without reference images — our villa tests show higher retry rates than on Kling for pure t2v facades.
- ▸Voice: use Runway Characters or external ElevenLabs — do not assume native lip-sync on Gen-4 video gens.
Runway Gen-4 vs Kling 3.0, Veo 3.1, Luma Ray 2
- ▸Wins: client-ready polish on vertical ads, image-to-video from brand stills, in-app fix tools without leaving Runway.
- ▸Loses on raw API cost at volume when retries stack — Kling Standard and Hailuo undercut per attempt.
- ▸Loses to Veo on speech-led UGC with native lip-sync in one pass.
- ▸Loses to Seedance on multi-reference brand campaigns.
Use case recommendations
Documentation verdict: Runway Gen-4 is the pick when the deliverable must look client-ready with minimal post — especially image-to-video social ads. Use Turbo for cost-sensitive iteration and Gen-4.5 when the client pays for the last increment of polish. See our real-estate and villa battles for logged side-by-side results.
Sources
- Runway Academy — Credits & available models ↗
- Runway — Developer API documentation ↗
- Runway — pricing page ↗
Want to see this model under pressure? Watch it battle with identical prompts →