GIF and SVG are GPT Image 2.6's two most talked-about new formats: one makes static images move, the other takes generated images into design workflows. But their maturity differs sharply — GIF works in first tests, SVG reliability is poor. This article lays out the parameters, limits and measured data.
In ChatGPT: a text prompt, a photo or a hand-drawn sketch can produce a looping GIF; playback speed is adjustable, frames are editable one by one; but you can't edit an existing GIF you upload; GIF generation is still rolling out gradually, and not all accounts have it yet. API: output_format: "gif" on both the generation and editing endpoints; frame_count 2–16 frames (default 9), frame_duration_ms 20–2,000ms per frame (default 200ms, rounded to the nearest 10ms); the default combination is 9 × 200ms = a 1.8s loop; GIF preview access is required; no streaming, partial images, custom compression or mask. Note the frame count: launch-day docs and early coverage said 2–64 frames; after Nov 7 the docs read 2–16 frames (default 9); kingy.ai published a correction on Nov 9. The current figure is 2–16.
In Analytics Vidhya's five hands-on tests on Nov 5, the ChatGPT browser test produced a genuine 12-frame GIF, a 2.16s loop. Syntx's two-day hands-on reached a similar conclusion: clean loop seams and stable subject size across frames. OmniaKey requested 9 × 200ms on Nov 5 and got 9 × 180ms, 1.62s total — a slight frame-duration discrepancy. OpenAI hasn't published GIF duration, frame-rate or file-size caps; the longest animation is bounded by the 16-frame limit and the total-pixel budget.
The guide is explicit about the two-stage flow: the API first generates a reference image (quality controls that image), then produces the SVG from it; the svg.model field names the author model that writes the vector code; billing adds the two stages, with no fixed per-image price. API-only (no SVG export in ChatGPT, per the system card); set output_format: "svg" on the generation endpoint, n must be 1; the editing endpoint doesn't support SVG output, and SVGs can't be used as editing input. The result is static vector content: editable paths, shapes and colors; no scripts, external resources or embedded bitmaps; text stays live editable text — CellCog notes that Recraft outlines each letter into shapes, while 2.6's text is live text.
CellCog's self-funded test on Nov 4–5: only 9 of 48 valid requests returned an SVG (19%), all failures bare HTTP 502s (after 11–91s); prompts with text succeeded 3 of 41, prompts without text 6 of 7; successful calls median 20.9s. The same batch on Recraft V4.1 was 9/9, median 8.3s, fixed $0.08/image. CellCog measured $0.03–$0.23 per returned SVG. CellCog says it re-runs the same batch weekly — no updated result as of Nov 14. The system card also notes a GIF safety risk: safeguards may miss violations involving motion or cross-frame meaning.
GIF: good for social-media animations, small product demos, stickers; the default 1.8s loop is enough for most uses, and for longer clips raise frames to 16 and frame duration to 2,000ms (the theoretical ceiling is ~32s, but the total-pixel budget limits actual length). SVG: for now, experiments only when you're not in a hurry — a 19% success rate and 502 failures mean production workflows can't depend on it; for reliable vector output, use Recraft V4.1 (9/9, fixed $0.08/image) or wait for CellCog's weekly re-runs. For the API parameter details, see our API access article.
Reference: CellCog — Recraft V4.1 vs GPT Image 2.6 SVG (rewritten and organized by ModelBrief; facts per the source)