ByteDance’s Seedance 2.5 is commercially available on multiple platforms with active 4K tiers and public pricing, while Meta Muse Video remains restricted to a preview status without a release date, public price, or resolution details since Meta Superintelligence Labs introduced it on July 7, 2026.
Seedance 2.5 Offers Platform Access and Competitive Benchmarks
- Seedance 2.5 is fully accessible across commercial platforms like Runway and Picsart, whereas Meta Muse Video lacks an endpoint or sign-up window.
- Artificial Analysis benchmarks rank Seedance models competitively on text-to-video, image-to-video, and editing boards, while Meta Muse Video does not appear on public evaluation boards.
- Both labs note identical technical hurdles—specifically complex rapid motion and multi-subject scene interaction—representing the current limitations of video diffusion models.
The Marketplace Reality of Seedance 2.5
ByteDance released Seedance 2.5 on July 31, 2026, establishing a public rate card and deploying a 4K tier across commercial platforms. OrcaRouter reported that availability includes BytePlus ModelArk under model ID dreamina-seedance-2-5-260628, Jimeng AI, Doubao Pro, the BytePlus playground, Dola, and Runway and Picsart. The model delivers clips up to 30 seconds in a single pass with multi-day extension capabilities.
Pricing for Seedance 2.5 stands at $11.70 per million video tokens, translating to roughly $0.57 per second at 1080p based on BytePlus examples, which Artificial Analysis normalizes to $34.12 per minute. Resolution spans 480p, 720p, and 10-bit 1080p through its native API, alongside invitation-only Volcano Engine trials.
Evaluating the Meta Muse Video Preview
Meta Superintelligence Labs introduced Meta Muse Video on July 7, 2026, describing it as coming soon for creators and Meta AI. Four months later, the system features no general availability date, price, clip duration, or resolution. OrcaRouter said that the preview post claimed competitive performance regarding prompt adherence, visual fidelity, and temporal consistency.
During Connect, we previewed Muse Realtime Avatar, our real-time embodiment technology that turns Muse Realtime Voice into expressive, interactive avatars. Today, we’re rolling it out in the @Muse app in the US, giving people a more natural way to engage with their personal… pic.twitter.com/QYU6Pg8Fch
The same post placed the system at number three in human-preference Elo for text-to-video on Arena as of July 5, 2026. However, that placement relies entirely on vendor-reported preference metrics rather than measured capability scores. The announcement acknowledged active technical gaps, stating that audio-video synchronization and physically accurate fast motion are still being developed.
Seedance Models Rank Highly on Artificial Analysis Scoreboards
Artificial Analysis scoreboards position Seedance models across multiple test boards. Seedance 2.5 holds the fourth spot on the T2V v2.1 board with an Elo of 1,155, while securing first place on the I2V v2.1 image-to-video board with an Elo of 1,161. On video editing tasks, it reaches fourth place on the editing board with an Elo of 1,147.
Comparing the disclosures from both labs reveals a shared technical boundary. ByteDance points out that complex motion physics and multi-subject interaction require further refinement. Meta identifies synchronized audio timing and rapid movement as active development challenges.
Production Workflows While Waiting for Muse
OrcaRouter does not route Meta Muse Video due to its lack of an API endpoint, though it serves Meta Muse Spark 1.2 at list prices of $1.25 per million input tokens and $4.25 per million output tokens for text workflows. For video generation, MiniMax H3 operates as an available alternative at $0.08 per second, holding an established position on text-to-video leaderboards.
