Gemini Omni Flash 1,240 vs Seedance 2.0 1,225: what 15 Elo means

Hedra's ranking, read 2026-10-03, lists Gemini Omni Flash first for text to video with audio at an Elo of 1,240, Seedance 2.0 4K second at 1,225 and Happy Horse 1.1 fourth. A 15 point Elo gap means the higher model is expected to win a head-to-head vote about 52 percent of the time, which is close to a coin flip. These are Hedra's cited numbers, not Sume's.

The ranking

Hedra's page lists these positions; only the first two Elo values appear in the source read.

What 15 points means

Elo converts a rating gap to an expected win rate with 1 / (1 + 10^(-gap/400)). For 15 points that is about 0.52. A gap that small is well inside the noise of prompt choice, so use the ranking to build a shortlist, not to pick a winner.

Limits that decide the pick

Sume's docs give Gemini Omni Flash 1.1 (gemini-omni-flash-1.1) a range of 3-10 seconds at 360p, 720p, 1080p and 4K, in 16:9 or 9:16, with native synced audio always on. Seedance 2.5 accepts 4-30 seconds at 480p, 720p and 1080p, and seedance-2 also offers 1080p. If your clip must be 20 seconds, the Omni range rules it out before Elo matters.

Omni also takes video edits through a video_url field and up to 10 reference images; the edit source cannot be combined with image_url, end_image_url or reference URLs.

What Elo does not tell you

Elo ranks preference on the prompts voters saw. It says nothing about your aspect ratio, your maximum duration, whether you need reference audio, or what a clip costs. Two models that tie on Elo can differ by a large factor on price.

Before you choose, write down your hard requirements: duration, aspect ratio, resolution, audio behavior and budget per clip. Cross out any model that fails one. Elo breaks ties among what remains.

A shortlist method

Pick the top three models that meet your duration, resolution and audio needs. Run the same ten prompts on each and score them blind. Record cost per finished second from usage.cost; Sume bills provider list times 1.25. Elo is a public preference vote on generic prompts, and your product shots may rank differently.