ChatGPT Now Makes Animated GIFs With Images 2.6. What About Arabic Text?

ChatGPT can now make animated GIFs. OpenAI announced Images 2.6 on 4 November 2026, and it is live globally in ChatGPT and Codex on every subscription tier, Free included, on web, desktop and mobile. Developers get two new API models alongside it. OpenAI says the new version generates images 50% faster than before, and The Decoder notes that people now make more than 4 billion images a week across ChatGPT, Codex and the API, a billion more than in September.

What OpenAI has not published is how many GIFs or images each plan gets. That matters more than usual: from December, ChatGPT and Codex will share one usage budget, and OpenAI has not said how image generation counts against it.

Animated GIFs and Frame Editing in ChatGPT

ChatGPT users can now create animated GIFs from text prompts or uploaded photos, even a rough hand drawing: OpenAI showed a pen sketch of an owl on lined paper turned into an animated character. It is a long way from the still-image days of DALL-E 3. Prompts like “Can you make a gif of a cat dancing?” produce slight variations strung together flipbook-style. As reported by The Decoder (crediting The Verge), the model keeps the character recognisable from frame to frame, which earlier models struggled to do. Users can adjust animation speed, edit individual frames, and share.

Safety safeguards apply specifically to motion. OpenAI’s system card states some content categories are blocked more broadly in GIFs than in still images because of inferred motion. OpenAI’s image generation guide also notes the model “can still struggle with precise text placement and clarity,” complex prompts take up to 2 minutes, and multi-generation consistency may struggle.

API Features: SVG Vectors and Token Pricing

In the API, OpenAI launched two models: gpt-image-2.6-flare and gpt-image-2.6-sunburst. Flare at quality=low returns images in as little as six seconds, while Sunburst offers higher quality and consistency across edits at medium and high settings. OpenAI’s image generation guide details new capabilities:

• GIF Output: Supported on /v1/images/generations with frame_count 2-64 (default 9) and frame_duration_ms 20-2000 (default 200, creating a 1.8-second loop). The edit endpoint does not output GIF.
• SVG Output: Supports editable paths for icons and logos (API only, generations only, n=1, no streaming). The edit endpoint does not support SVG.
• Complexity and Markup: A new complexity field (low, medium, high; default high) sets how much the model plans before rendering, at no extra fee: low suits quick edits such as swapping a background, high suits dense text and busy layouts. Image markup, where you mark the exact area to change, moves from ChatGPT to the API.

Prices on OpenAI’s pricing page match GPT-Image-2.5: image input $8 / 1M tokens ($2 cached), image output $30 / 1M tokens, and text input $5 ($1.25 cached). However, OpenAI publishes no per-image price for 2.6, which MarkTechPost criticized. The Batch table also did not list 2.6 models when checked on 5 Nov 2026.

The Arabic Script Test: What to Check

OpenAI says nothing about Arabic, non-Latin scripts, or right-to-left text, and we did not test it. Run a test in ChatGPT: request a short looping GIF with one Arabic word (such as “مرحبا” or “عيد سعيد”) on a sign, and check three points:

• Letter Joining: Confirm letters join correctly rather than rendering as disconnected characters.
• Right-to-Left Direction: Verify characters read in the proper right-to-left order.
• Frame Drift: Check that the word stays identical across all frames, as separate frame generation can cause drift.

For practical workflows, generate textless GIFs in ChatGPT and add Arabic typography afterward in Canva, the same advice we gave for Canva’s Craft image model. For SVG logos via the API, generate the symbol and set the Arabic wordmark yourself. ChatGPT access is global wherever the service operates; API prices are listed in US dollars, with no regional pricing on the page.

What to Do Now: Who Actually Benefits?

Images 2.6 delivers real value for creators needing quick animated GIFs without editing software, and developers gaining clean SVG vector output. Teams budgeting at scale will find token-only pricing, with no per-image rate, harder to plan around. Arabic-first brands have no assurance at all: OpenAI’s own guide admits text placement can still go wrong and never mentions Arabic. Until your own test says otherwise, use Images 2.6 for motion and drafts, and set the Arabic lettering yourself.