Ideogram 4.0

Overview

Ideogram 4.0 is a 9.3 billion parameter open-weight text-to-image model from Ideogram, released on June 3, 2026. It leads the open-weight field at its size on in-image text rendering, scoring 0.97 on the X-Omni English OCR benchmark, and pairs that with bounding-box layout control, structured JSON prompting, color palette conditioning, and output up to 2K. The weights, inference code, and prompting guide are public, with quantized builds that run on a single 24 GB GPU.

What Ideogram 4.0 does differently

Most image models take a sentence and return pixels, so text comes out misspelled and placement is a roll of the dice. Ideogram 4.0 was trained exclusively on structured JSON captions, so a prompt can spell out each element: where it sits in the frame, how it is styled, the exact string a text element should render, and the hex colors the image must stay inside. That structure makes results precise and repeatable, which matters for design work that ships as posters, packaging, banners, and UI rather than one-off art.

Ideogram 4.0 and Morphic

On Morphic, Ideogram 4.0 has been replaced by its successor, Ideogram 4.5, which keeps the same strength in in-image text and adds precise, repeatable editing. Select Ideogram 4.5 in Copilot or on the Canvas and it runs alongside the rest of the image line, plus video, speech, and music, in one workspace. To compare AI image models on text accuracy and maximum resolution, see how the Ideogram line stacks up against Nano Banana 2, GPT Image 2.5, and Recraft V4.1 Pro.

Ideogram 4.5 on Morphic generates at nine aspect ratios, 9:16, 2:3, 3:4, 4:5, 1:1, 5:4, 4:3, 3:2, and 16:9, at 1080p or 1440p. Reach for it when the words inside the image have to be spelled correctly and set well. There is a dedicated Ideogram 4.5 image editor if you want to start from an image you already have.