gpt-image-1
Ranked 2 of 22 image generation models by recommenders
People recommend it for following detailed prompts, careful localized edits and fun personal image tasks. The most common caveat is that it loses fine details and hallucinates, and it is slow and costly per image.
People recommend it for following detailed prompts, careful localized edits and fun personal image tasks. The most common caveat is that it loses fine details and hallucinates, and it is slow and costly per image.
158 people gave an opinion
- recommend it
- 79
- criticize it
- 79
- approval share
- 50%
Recommended in 12 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 79 people recommend itNeeds 10 or more, met.
- 1 recommender for every criticNeeds 2 to 1 or better, not met.
- Recommended in 12 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Follows detailed prompts closely
“SOTA models like NB Pro and gpt-image-1 are really good at adhering to even moderately complicated prompts”
Comment from 12 Feb 2026 in “Show HN: Image prompt game with multi-signal CLIP/HSV/HOG scoring” “Following the prompt instructions more closely also seems better compared to Nano Banana Pro.”
Comment from 17 Dec 2025 in “GPT Image 1.5” “gpt-image-1 had the famous "piss filter"”
Comment from 1 Mar 2026 in “Show HN: Imagedojo.ai – Blind arena for Google, OpenAI, and xAI image generators” Careful localized image edits
“It's leagues better at making localized edits without altering the entire image's aesthetic than gpt-image-1”
Comment from 16 Dec 2025 in “GPT Image 1.5” “GPT-Image 1.5 is clearly a step up as an editing model, but it performed worse in purely generative tasks”
Comment from 17 Dec 2025 in “GPT Image 1.5” “It's absolutely OP at understanding how to turn low-fidelity renders into final draft upscales.”
Comment from 30 Jan 2026 in “Antirender: remove the glossy shine on architectural renderings” - Comment from 16 Dec 2025 in “GPT Image 1.5”
Fun for personal everyday images
“Now I can just generate illustrations that fit my use case well, and it takes very little effort.”
Comment from 9 Sep 2026 in “ChatGPT Images 2.5” “Having a lot of fun with my son using an image of him and dressing him in all kinds of costumes”
Comment from 8 Sep 2026 in “ChatGPT Images 2.5” “You can take the photo, inside the chatgpt app and then talk to it without much effort.”
Comment from 9 Sep 2026 in “ChatGPT Images 2.5” - Comment from 8 Sep 2026 in “ChatGPT Images 2.5”
Strong creative style-heavy edits
“Best for creative, transformative, style-heavy edits, but it hallucinates a lot and often distorts fine details”
Comment from 12 Nov 2025 in “We ran over 600 image generations to compare AI image models” “gpt-image-1 absolutely excels at this, though you don't have much control over the style and aesthetic”
Comment from 16 Dec 2025 in “GPT Image 1.5”
What people warn about
Each mark is one comment
Loses and distorts fine details
“Best for creative, transformative, style-heavy edits, but it hallucinates a lot and often distorts fine details”
Comment from 12 Nov 2025 in “We ran over 600 image generations to compare AI image models” “still the miniscule details are being lost, like the teeth structure of the guy in the middle”
Comment from 8 Sep 2026 in “ChatGPT Images 2.5” Slow and expensive per image
“The latency is still too high, lower than 10 seconds for nano banana and around 25 seconds for GPT image 1.5”
Comment from 17 Dec 2025 in “GPT Image 1.5” “gpt-image 2.5 being great at times, at almost 1 order of magnitude more of the cost per image and way slower”
Comment from 15 Oct 2026 in “Krea 3 – An image model built for editing at scale” Often used for fake or misleading images
“the examples in OpenAI’s blog post are examples of creating images that have no plausible purpose other than to fake stuff that didn’t happen”
Comment from 8 Sep 2026 in “ChatGPT Images 2.5” “The only use case I can think of for this is for someone who likes to make up stories and lie about what they’ve done.”
Comment from 8 Sep 2026 in “ChatGPT Images 2.5” “All I see people using this for is generating slop for work presentations, and misleading people on social media.”
Comment from 17 Dec 2025 in “GPT Image 1.5” - Comment from 8 Sep 2026 in “ChatGPT Images 2.5”
- Comment from 30 Apr 2026 in “Accurate infographics with ChatGPT Images 2”
Weaker than rivals in consistency
“Even from their own curated examples, this looks quite a bit worse than nano banan in terms of preserving consistency on image edits.”
Comment from 16 Dec 2025 in “GPT Image 1.5” “GPT-Image 1.5 is clearly a step up as an editing model, but it performed worse in purely generative tasks”
Comment from 17 Dec 2025 in “GPT Image 1.5”
Other image generation models people recommend
The four most recommended other image generation models, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| Nano Banana | 265 | 159 | 63% | 12 | 6 mo5 owners | — |
| gpt-image-1This page | 79 | 79 | 50% | 12 | — | — |
| FLUX.2 | 66 | 51 | 56% | 10 | — | — |
| gpt-image-2 | 63 | 26 | 71% | 8 | — | Approved |
| Qwen-Image | 47 | 38 | 55% | 10 | — | — |
What people said
40 of the 211 comments that mention it, newest first. Each links to the original comment.
“gpt-image 2.5 being great at times, at almost 1 order of magnitude more of the cost per image and way slower”
“Now I can just generate illustrations that fit my use case well, and it takes very little effort.”
“You can take the photo, inside the chatgpt app and then talk to it without much effort.”
“the biggest uses of these sort of models is to pump out lots of content on multimedia-based social media”
“the examples in OpenAI’s blog post are examples of creating images that have no plausible purpose other than to fake stuff that didn’t happen”
“I used it to visualise different arrangements of my furniture and it worked nicely.”
“If this can allow them to clearly communicate prices in a well designed pleasant way, it's great.”
“The only use case I can think of for this is for someone who likes to make up stories and lie about what they’ve done.”
“ChatGPT Images 2.5 seems like an even better starting point to test out what these models can actually do nowadays”
“Having a lot of fun with my son using an image of him and dressing him in all kinds of costumes”
“This is awesome and is a direct consequence of everything I ever wanted the future to be: magical creative superpowers.”
“ChatGPT Images is for me the most impresive of all models, but also the one I hate the most”
“still the miniscule details are being lost, like the teeth structure of the guy in the middle”
“ChatGPT detail is a lot better though. You can do stuff like complex 6-panel comics that Nano Banana can't match.”
“I remembering being absolutely shocked when the gpt-image series managed to pass the Labyrinth test”
“after a few more additions like spotted fur and multicolored tentacles, it was no contest”
“Images 2.0 being the poster child of this "agentic iterative image composition" approach.”
“it presented the image as if it were using up-to-date data but it clearly pulled the records from 2025”
“OpenAI's models have always been very strong in prompt adherence but visually weak”
“gpt-image-1 had the famous "piss filter"”
“SOTA models like NB Pro and gpt-image-1 are really good at adhering to even moderately complicated prompts”
“The first is very fake and the second is a strong improvement, though still far from the excellent cafe shots above”
“They release GPT Image which honestly is probably the main reason they still have users they have.”
“It's absolutely OP at understanding how to turn low-fidelity renders into final draft upscales.”
“GPT-image-1 (and gpt-image-1.5) are outliers in their ability to generate images with a proper alpha channel.”
“Out of curiosity why does gemini get gold for the poker example but gpt-image 1.5 does not?”
“GPT-Image seems to trade a bit of micro-detail for better global coherence, especially in lighting”
“It moved some of the marks around to accommodate the enlarged black area, but retained the overall appearance of the sign.”
“The latency is still too high, lower than 10 seconds for nano banana and around 25 seconds for GPT image 1.5”
“This outperforms Gemini 3 pro image (nano banana pro) on Text-to-Image Arena and Image Edit Arena.”
“GPT-Image 1.5 is clearly a step up as an editing model, but it performed worse in purely generative tasks”
“All I see people using this for is generating slop for work presentations, and misleading people on social media.”
“Following the prompt instructions more closely also seems better compared to Nano Banana Pro.”
“gpt-image-1 absolutely excels at this, though you don't have much control over the style and aes”
“gpt-image-1 absolutely excels at this, though you don't have much control over the style and aesthetic”
“It's leagues better at making localized edits without altering the entire image's aesthetic than gpt-image-1”
“Gpt-image-1, on the other hand, understands the layout and blocking of the scene, the pose of human characters, and will literally repair and upscale everything.”
“we seriously can't be burning GW of energy just to have sama in a GPT-Shirt Ad generated by A.I”
“Even from their own curated examples, this looks quite a bit worse than nano banan in terms of preserving consistency on image edits.”
“Best for creative, transformative, style-heavy edits, but it hallucinates a lot and often distorts fine details”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.