GPT Image 2.6 errors: 'No supported sheet fits GIF frames' and svg.model fixes
Fix GPT Image 2.6 errors for GIF and SVG output: the GIF sheet size limit, invalid frame_count and frame_duration_ms, the missing svg.model and n=1 for SVG. Covers complexity and the other new parameters.
Ofox 
TL;DR: Most first-day GPT Image 2.6 errors come from the new GIF and SVG limits: 1024x1024 GIF frames do not fit on a sheet, frame_count must be 2–16, and SVG needs svg.model and n=1. Rules come from OpenAI’s API specification of November 4–5, 2026; error messages are OpenAI’s responses to our requests or quoted from its developer forum.
Applies to: gpt-image-2.6-sunburst and gpt-image-2.6-flare on the Images API (/v1/images/generations, /v1/images/edits), plus the image_generation tool in the Responses API. Last checked: 2026-11-05.
| You see | Go to |
|---|---|
No supported sheet fits 9 GIF frames at size '1024x1024' | GIF sheet size |
Invalid 'frame_count' for GIF output. or Invalid 'frame_duration_ms' | frame_count and frame_duration_ms |
Complexity 'low' is not supported for this image model. | complexity in the Responses API |
SVG generation requires 'svg.model'. | svg.model |
The 'n' parameter is not supported for SVG generation. Use n=1. | n=1 for SVG |
For how-tos, see the GIF tutorial, editable SVG guide and API pricing.
What is new in the GPT Image 2.6 API
For annotation-file preparation, use the markup editing file and request checks. If the request succeeded but an earlier edit disappeared or product details changed, follow the repeated-edit version and acceptance workflow; an output-quality problem is not itself one of the parameter errors below.
GPT Image 2.6 ships as two models: gpt-image-2.6-sunburst for quality and consistent edits, and gpt-image-2.6-flare for speed. Both accept the parameters below on top of the GPT Image 2.5 set.
| Parameter | Values | Rule from the spec |
|---|---|---|
complexity | low, medium, high | Planning effort, separate from quality. The model default applies when omitted. |
output_format | adds gif, svg | gif on generations and edits; svg on generations only. |
frame_count | 2–16, default 9 | Only with output_format: "gif". |
frame_duration_ms | 20–2000, default 200 | Only with GIF; rounded to 10 ms. |
svg | {model, reasoning} | Picks the model that writes the SVG. |
markup (edits) | transparent PNG | Supported on 2.5 and 2.6; cannot be combined with mask. |
GIF needs GIF preview access for the selected model. Neither GIF nor SVG supports streaming, partial images or output_compression.
No code needed: in Ofox Chat, ask “Use GPT Image 2.6 to draw …” and choose Flare or Sunburst. Chat returns PNG or JPEG; use the API for GIFs.
No supported sheet fits GIF frames
No supported sheet fits 9 GIF frames at size '1024x1024'. Invalid size '3072x3072'. Requested resolution exceeds the current pixel budget.GPT Image 2.6 returns the No supported sheet fits error (image_generation_user_error) when the GIF frames, laid out on one sheet, exceed the API limit. The numbers and the closing reason vary with your request:
No supported sheet fits 9 GIF frames at size '1536x1024'. Invalid size '4608x3072'. The longest edge must be less than or equal to 3840.No supported sheet fits 2 GIF frames at size '1024x1024'. Invalid size '4096x2048'. The longest edge must be less than or equal to 3840. Invalid size '2048x4096'. The longest edge must be less than or equal to 3840.Lowering frame_count does not help at 1024x1024: 2, 3, 4 and 9 frames all failed in our tests, because even the smallest sheet is 4096 pixels on its long edge.
- Request 512x512 frames:
{"model": "gpt-image-2.6-flare", "output_format": "gif", "size": "512x512", "quality": "low", "frame_count": 4, "prompt": "A small orange fox mascot waving"} - If you need a larger animation, upscale the GIF after generation.
How to confirm it worked: the response has "output_format": "gif" and the decoded file starts with GIF89a. At 512x512 we got 4 frames in 247 KB, 9 frames in about 480 KB and 16 frames in 977 KB.
If it still fails: remove partial_images, stream, mask and output_compression.
Docs: the Images API reference with request and response fields lists every parameter this endpoint accepts.
Invalid frame_count or frame_duration_ms for GIF output
Invalid 'frame_count' for GIF output. Received 40. Use an integer from 2 to 16, or omit 'frame_count' to use 9.Invalid 'frame_duration_ms': integer below minimum value. Expected a value >= 20, but got 5 instead.The frame_count and frame_duration_ms errors mean the value is out of range. OpenAI’s GIF guide lists frame_count as 2–64, but the API reference and the endpoint enforce 16: a developer forum report shows 17, 20, 32 and 64 frames all failing while 16 works.
- Set
frame_countto an integer from 2 to 16, or omit it for 9 frames. - Set
frame_duration_msto a value from 20 to 2000, or omit it for 200 ms. - Send both only with
output_format: "gif".
How to confirm it worked: open the GIF in a tool that shows frames. Our request with frame_count: 4 and frame_duration_ms: 100 returned 4 frames of 100 ms each.
Complexity ‘low’ is not supported for this image model
Complexity 'low' is not supported for this image model.After the 2.6 launch, developers reported Complexity 'low' is not supported for this image model (image_generation_user_error, param: complexity, code: invalid_value) from the Responses API image_generation tool with gpt-image-2, even though they never sent complexity. According to a developer forum thread, the outer reasoning.effort is passed to the image tool as complexity, and gpt-image-2 does not accept it; complexity applies only to GPT Image 2.6. Other users confirmed the behavior and said it was escalated.
- Remove
reasoning.effortfrom the Responses request if you do not need it. - If a router or SDK always adds it, set
"effort": null, the workaround from the thread. - Send
complexityonly togpt-image-2.6-sunburstorgpt-image-2.6-flare.
How to confirm it worked: the Responses call returns an image_generation_call item instead of the 400 error.
SVG generation requires svg.model
SVG generation requires 'svg.model'.GPT Image 2.6 returns SVG generation requires 'svg.model' (invalid_request_error) when output_format is svg and svg.model is missing. The image model draws a reference image, then a second model, the SVG author, writes the SVG code. The specification says svg.model is required when SVG model selection is enabled for the account; otherwise the default SVG author is used.
- Add the
svgobject and keep a single image:{"model": "gpt-image-2.6-sunburst", "output_format": "svg", "n": 1, "prompt": "Flat orange fox head logo", "svg": {"model": "<SVG author model>", "reasoning": {"effort": "medium"}}} - Use the top-level
modelfor the image model andsvg.modelfor the SVG author.svg.reasoning.effortis optional; supported values depend on the author model.
How to confirm it worked: the response includes svg_usage with the SVG author’s model name and token counts.
If it still fails: check that your account has SVG output and SVG model selection enabled, and log the full error body; the guide to API error codes and retries shows which errors are worth retrying.
n is not supported for SVG generation
The 'n' parameter is not supported for SVG generation. Use n=1.GPT Image 2.6 accepts only one SVG per request. Set n to 1, or remove it, and send one request per SVG.
How to confirm it worked: the error disappears and data contains one item.
Related Docs
- GPT Image models on OfoxAI: model IDs and when to use each
- Images API reference: generations and edits
Sources
- OpenAI, Introducing Images 2.6, November 4, 2026
- OpenAI API specification, commit 5882fdd: GIF and SVG support and new image parameters, November 5, 2026
- OpenAI Developer Community, Responses API with gpt-image-2 gives Complexity ‘low’ is not supported, November 5, 2026
- OpenAI Developer Community, GPT Image 2.6 GIF frame limit: guide says 64, API enforces 16, November 5, 2026
- OfoxAI test requests on November 5, 2026 (error messages above)
Frequently Asked Questions
- Why does GPT Image 2.6 say “No supported sheet fits GIF frames”?
- GIF frames are rendered together on one sheet, and the sheet is larger than the API allows. In our tests 1024x1024 failed at every frame count we tried, from 2 to 9, while 512x512 returned GIFs of up to 16 frames.
- How do I fix “SVG generation requires 'svg.model'”?
- Add an svg object with a model field naming the model that should write the SVG, and keep n at 1. OpenAI's specification says svg.model is required when SVG model selection is enabled for the account.
- What values do frame_count and frame_duration_ms accept?
- frame_count takes an integer from 2 to 16 and defaults to 9. frame_duration_ms takes 20 to 2000 milliseconds, defaults to 200 and is rounded to 10 ms. Both work only with output_format gif.
