model release

Ideogram 4.5 launches with native 2K output and four tiers from $0.008 to $0.22 per image

TL;DR

Ideogram has released Ideogram 4.5, an image model it claims edits only the area a user specifies and leaves the rest untouched. It offers four quality tiers from 0.8 to 22 cents per image, all at native 2K resolution, via the Ideogram platform and API. An open-weight release is promised but not yet dated.

2 min read
0

Ideogram has released Ideogram 4.5, an image model that the company claims modifies only the region a user specifies and leaves the rest of the image unchanged. The model is available now on the Ideogram platform and through the API, with four quality tiers priced from $0.008 to $0.22 per image, all at native 2K resolution.

Key details

  • Pricing: Four quality tiers, from 0.8 cents to 22 cents per image. Tier names and the exact price of each tier were not specified in the source report.
  • Resolution: Native 2K across all tiers.
  • Availability: Ideogram platform and API, live now.
  • Partners: Picsart, Runway, Pika, and Leonardo AI.
  • Open weights: Ideogram says an open-weight release is coming soon. No date, license, or model size has been disclosed.
  • Not disclosed: Parameter count, training cutoff, and benchmark scores. A context window is not applicable to the source report's description of the model.

The problem Ideogram is targeting

Localized editing is a persistent weakness in image generation. When a user changes one element, such as swapping a person's outfit, the subject's body shape and the background should stay intact. According to the report, models such as GPT-Image 2.5 and Nano Banana have improved considerably here but still tend to produce artifacts, particularly after several consecutive edits.

Ideogram claims 4.5 "only touches what the user tells it to." The company lists product photography, interior design, photo restoration, and in-image text editing as target use cases.

What is verified and what is not

The consistency claim comes from Ideogram. The Decoder reports that early testers confirm strong consistency across edits, but no named testers, test methodology, or quantitative results were published. Ideogram has not released head-to-head benchmark numbers against GPT-Image 2.5 or Nano Banana, so the comparison remains unquantified.

The partner list suggests immediate distribution beyond Ideogram's own product. Picsart, Runway, Pika, and Leonardo AI are all established creative-tool platforms. Whether each partner has already integrated 4.5 or will do so later was not specified.

What this means

Edit stability across multiple passes matters more than single-shot quality for professional workflows. Product photography and interior design rely on iterative changes, and cumulative drift forces users to restart or fix results manually. If Ideogram's claim holds up under independent testing, 4.5 addresses a real barrier to production use.

The pricing spread is also notable. A 27.5x gap between the cheapest and most expensive tier lets developers use low-cost tiers for drafts and reserve the top tier for final output. Without tier-level detail, though, teams cannot yet model costs accurately.

Two things will determine the model's standing: independent evaluations of multi-edit consistency, and the open-weight release. Ideogram has given no timeline or license terms for the weights, and those details will decide whether the release matters to self-hosting developers and the research community.

Related Articles

model release

inclusionAI releases Duo 1.0 causal-base, a full-duplex multimodal world model initialized from MiniMax H3

inclusionAI has published the Duo 1.0 causal-base weights on Hugging Face. Duo is described as a full-duplex multimodal world model for real-time audio-visual interaction, initialized from MiniMax H3 and trained further by the Duo team.

model release

Perplexity releases Decider v1.3 and Decider Flash v1.3 with text-span extraction and Apache-2.0 weights

Perplexity released Decider v1.3 and Decider Flash v1.3 on November 12, 2026, adding verbatim text-span extraction and downloadable Apache-2.0 weights. The company reports 91.6% mean accuracy for the larger model across 63 query-classification tasks.

model release

Huawei Cloud releases CloudRobo-WM1.2 world action model, claims first place on RoboTwin 2.0 and LIBERO-Pro 2.0

Huawei Cloud has released CloudRobo-WM1.2, an embodied world action model, and says it will be open source. The company claims first place on the RoboTwin 2.0 and LIBERO-Pro 2.0 robot manipulation benchmarks, with scores of 88.98% and 81.7%.

model release

Tencent launches Hy4 flash at $0.07 per 1M input tokens, one-twentieth the per-task cost of Hy4

Tencent launched Hy4 flash on Monday, a lower-cost model for everyday work such as research, slide creation and spreadsheet editing. On Tencent's internal WorkIndex benchmark, it cost one-twentieth as much per task as the full Hy4 model. API list prices are $0.07 per million input tokens and $0.281 per million output tokens.

Comments

Loading...