Runway
The strongest independent AI media stack — frontier video plus a real developer platform — but it's a media specialist, not a full AI stack, and credits meter everything.
Element scores
Strengths
Gen-4.5 topped the Video Arena leaderboard ahead of Google and OpenAI (12/2025) and now ships native audio and multi-shot generation; Aleph 2.0 and Edit Studio brought frame-level in-video editing (05/2026); GWM-1 world models (Worlds, Avatars, Robotics) opened real-time simulation and synthetic-data product lines (12/2025). Runway Dev (08/07/2026) turns it all into a one-integration developer platform — first-party frontier models plus day-zero third-party models (Veo 3.1, Seedance 2.0, Kling 3.0, Nano Banana Pro), Recipes, Workflows and interactive Characters — already used by Adobe, ElevenLabs, Shutterstock, Figma and Gamma. $315M Series E (02/2026) at a $5.3B valuation plus IMAX and AMC Networks partnerships give it staying power.
Honest dings
Credit metering is everywhere and hard to reason about across dozens of models with different burn rates — Standard's 625 credits buy roughly 52 seconds of Gen-4.5 a month, so serious use means Pro or Max. No eval/QC tooling ships despite API-scale positioning. The agent-era surfaces (Characters tool calling, knowledge grounding) are brand-new and thin, and there is no MCP support. Aggregating rivals' models is convenient but means the best model on the platform is frequently not Runway's own.
Sources
Every audit lists the research it rests on — transparency and traceability are the product. Tools evolve: each audit is a snapshot of its audit date, and re-audits supersede older versions (kept below for reference).
- runwayml.com/pricing — Official live pricing: Free/Standard/Pro/Max/Enterprise tiers, per-model credit costs, 50M+ creators claim (accessed 2026-07-12)
- runwayml.com/news/introducing-runway-dev — Official Runway Dev launch (08/07/2026): Models/Recipes/Workflows/Characters, enterprise customers (Adobe, ElevenLabs, Shutterstock, Figma, Gamma), zero-data-retention commitments (accessed 2026-07-12)
- docs.dev.runwayml.com/guides/using-the-api — Official API docs: Gen-4.5 quickstart, SDKs, usage tiers, Agent Skills (accessed 2026-07-12)
- github.com/runwayml/skills/blob/main/skills/rw-api-… — Official model/endpoint/credit table: gen4.5 (12 cr/s), gen4_turbo (5), gen4_aleph (15), act_two (5), third-party veo3/veo3.1/seedance2; input limits (accessed 2026-07-12)
- runwayml.com/research/introducing-runway-gwm-1 — Official GWM-1 announcement (11/12/2025): autoregressive world model on Gen-4.5, Worlds/Avatars/Robotics variants (accessed 2026-07-12)
- techcrunch.com/2025/12/11/runway-releases-its-first… — Independent: GWM-1 launch coverage; Gen-4.5 surpassed Google and OpenAI on the Video Arena leaderboard; GWM Robotics SDK (accessed 2026-07-12)
- arstechnica.com/ai/2025/12/with-gwm-1-family-of-wor… — Independent analysis: GWM-1 is three separate post-trained models, not one general model; use cases and caveats (accessed 2026-07-12)
- cnbc.com/2026/05/19/runway-cnbc-disruptor-50-rankin… — Independent: $315M Series E led by General Atlantic (02/2026) at $5.3B valuation; IMAX, AMC Networks, Fabula partnerships; Gen-4.5 milestone (accessed 2026-07-12)
- releasebot.io/updates/runwayai — Independent release tracker: Aleph 2.0 + Edit Studio (21/05/2026), Seedance 2.0 on API (05/2026), Seedance 2.0 Mini (25/06/2026), Seed Audio 1.0 — shows day-zero third-party cadence (accessed 2026-07-12)