Seedance 2.5 vs Wan 3.0

Seedance 2.5 vs Wan 3.0 at a glance

Both run on the same Revana API key; switching is the model field. Prices are Revana’s, per second of video, as of November 2026.

What a clip costs on each

The price of one clip with sound at each length, at 720p and 1080p. Where a model cannot make that length in one take, the cell shows what the same seconds cost in several clips.

Where Seedance 2.5 wins

• Up to 30 reference images, against 10 on Wan 3.0, plus reference clips and sounds.
• Minimum 4 seconds, up to 30: the same ceiling as Wan 3.0.

Where Wan 3.0 wins

• Much lower price. 480p $0.04 · 720p $0.07 · 1080p $0.14 a second, against 480p $0.09 · 720p $0.21 · 1080p $0.52 for Seedance 2.5. A 30 second 1080p clip is $4.20 on Wan 3.0 and $15.60 on Seedance 2.5.
• Shorter clips. From 2 seconds.
• A higher tier. Wan 3.0 Prime (480p $0.07 · 720p $0.13 · 1080p $0.26) when you want more from Wan at still less than Seedance 2.5.

Which to use

Wan 3.0 is the budget pick for long clips: start there, and move a shot to Seedance 2.5 when it needs more than 10 references, reference clips or sound references, or when Wan’s take does not hold up.

Test both with one prompt

Picture quality is a matter of the shot, so the honest test is your own prompt on both. Quote both first (free), then render each at its lowest resolution.

export REVANA_API_KEY="sr_live_..."
# Same prompt on both models at their lowest resolution. Quotes are free: check them first.
for PAIR in seedance-2.5:480p wan-3.0:480p; do
  MODEL=${PAIR%%:*}; RES=${PAIR#*:}
  curl -s https://api.revana.si/v1/generations/quote \
    -H "Authorization: Bearer $REVANA_API_KEY" -H "Content-Type: application/json" \
    -d '{"model": "'$MODEL'", "input": {"prompt": "A barista pours latte art in a sunlit cafe, slow push-in, steam rising, soft jazz and cups clinking", "duration": 5, "resolution": "'$RES'", "audio": true}}'
  echo
done
# Then create each one with the request in the API section of /seedance-2-5, and compare the two clips.

From Claude or ChatGPT, ask the Revana MCP server: “Quote this prompt on Seedance 2.5 and Wan 3.0 at the lowest resolution, then make both.”