I asked ``Create an image with Gemini,'' but for some reason only text was returned. No matter how many times I ask them to "generate an image," the text continues to explain...

As of February 2026, an increasing number of people are experiencing the problem of "images not appearing" with Google Gemini's image generation function. On X (formerly Twitter), there have been comments such as ``Gemini, even though I am instructed to generate an image, it keeps outputting text'' and ``This morning, all I get is an error.''

In this article, we will organize 5 reasons why images are not generated in Gemini and explain how to deal with each in an easy-to-understand manner. If you're wondering, "How can I get an image?", let's clear it up here.

Cause 1: The model being used does not support image generation

Actually, this is the most common cause. Gemini has multiple models andnot all models can create images.

As of February 2026, the following models are mainly compatible with image generation.

  • Gemini 2.0 Flash (powered by Imagen 3) — Gemini app default
  • Models available for Gemini Advanced

On the other hand, text-specific models such as Gemini 3.1 Pro Preview (released February 19, 2026) can "analyze" images but cannot "generate" them. In other words, if the model automatically switches to the text version, no matter how much you ask it to ``create an image,'' it will only return text.

Workaround:

  1. Check your model name at the top of the Gemini app screen
  2. Switch to image generation compatible model (such as Gemini 2.0 Flash) from the model selection menu
  3. If you are using Gemini Advanced (paid version), choose an image-compatible model explicitly in the model selection dropdown

Cause 2: You are stuck in the free version's number of times limit

Gemini's image generation function has a daily generation limit for each plan. According to Google official help page, the rough guideline as of February 2026 is as follows.

  • Free version: High-quality images (Imagen 3) are limited to a few images per day. Automatically switches to standard model when upper limit is exceeded
  • Google AI Plus (approximately 1,200 yen per month): approximately 50 images/day
  • Google AI Pro: Approximately 100 images/day
  • Google AI Ultra: Approximately 1,000 images/day

If you try to generate a lot of images with the free version, it may switch to a "text-only explanation" halfway through. Sometimes you'll get an error message, or sometimes the output will change to text without any explanation, making you wonder if it's broken.

Workaround:

  1. Wait a while and try again the next day (the upper limit is reset daily)
  2. If you frequently generate images, consider upgrading to Google AI Plus
  3. As an alternative, use ImageFX (a free image generation tool provided by Google)

Cause 3: Blocked by safety filter

Gemini is equipped with a very strict safety filter. Prompts like the one below may block image generation and only return a text response.

  • Prompt containing real person's name (e.g. "Photo-style illustration of 〇〇")
  • Copyrighted characters (e.g. "Pikachu style illustration")
  • Content that suggests violence or sexual expressions
  • Request for realistic depiction of historical figures

Roughly speaking, in 2024, Google received criticism for ``creating historically inaccurate images'' when generating images of people, and since then it has significantly strengthened its safety filters. Google official blog also announced that it would "work on improvements." In some cases, the generation of human images itself is restricted in the free version.

Workaround:

  1. Do not include the person's name, but describe them using their characteristics (e.g. "a 30-year-old businessperson wearing a suit")
  2. Avoid photo realism by specifying the style such as ``illustration style'' or ``flat design''
  3. Write the purpose at the beginning of the prompt (e.g. "For an eye-catching blog post...")
  4. If you really need images of people, try Gemini Advanced (paid version)

Cause 4: The prompt is too vague

Gemini is a smart AI, but it tends to respond to ambiguous instructions with text instead of images. For example, if you just say ``Create a cute image,'' the app may explain in text how to make a cute image.

The trick to creating image generation prompts is to be specific and structured.

Bad example:

  • “Images of beautiful scenery”
  • “Make a cat image”

Good example:

  • "Japanese garden with cherry blossoms in full bloom, landscape with cherry blossom petals floating in the pond, photographic style, horizontal"
  • "Brown tabby cat curled up on the sofa, warm indoor light, watercolor style illustration"

The key is to include the following four things.

  1. Subject (what to draw)
  2. Place/background (where)
  3. Style (photographic style, illustration style, watercolor style, etc.)
  4. Composition/atmosphere (horizontal, bright, retro, etc.)

Also, English may be more accurate than Japanese. If you're not good at English, an effective technique is to first write a prompt in Japanese and ask Gemini to translate this prompt into English and generate an image in that English.

Cause 5: Server failure or maintenance

If the Gemini server has a problem or is undergoing maintenance, only image generation may stop working. The pattern is that the text response is normal, but the image is not generated.

In February 2026, there were multiple reports on X such as ``Gemini regularly throwing out errors'' and ``conversation history completely disappearing.''

Workaround:

  1. Check the failure information on Google Cloud status page
  2. Check Gemini problem reports on Downdetector
  3. Search for "Gemini image generation" and "Gemini error" on X (old Twitter) to see if anyone has the same symptoms
  4. Wait 30 minutes to 1 hour and try again

Checklist if the image still does not appear

If the above five causes do not resolve the issue, please try the following in order.

  1. Delete your browser cache then try accessing Gemini again
  2. Try in a different browser (Chrome, Edge, Firefox, etc.)
  3. Try the same prompt in Smartphone app version of Gemini
  4. Check the age settings of your Google account (according to Google official policy, image generation is not available for those under 18 years of age)
  5. If you are connected to a VPN, try turning off the VPN
  6. Start a new chat and then ask for image generation (the context of the conversation may influence this)

FAQ

Can I use Gemini's image generation for free?

Yes, image generation is available in the free version as of February 2026. However, there is a limit to the number of images that can be generated per day, and only a few high-quality images (Imagen 3) can be generated. If you use it frequently, we recommend upgrading to Google AI Plus (approximately 1,200 yen per month).

Can I create images of people with the free version of Gemini?

The free version has major limitations in generating human images. Generations that specify the name of a real person will be blocked, and even fictitious people may fall through the safety filter. Although some restrictions are relaxed in the paid version (Gemini Advanced), it is not possible to generate completely freely.

Why do I get a text explanation when I say "generate an image"?

The most common cause is that "a model that does not support image generation is selected." Check the model name at the top of the Gemini screen and try switching to a model that supports image generation (such as Gemini 2.0 Flash). Text responses may also occur if the prompt is too vague.

Are there any tools that can generate images for free instead of Gemini?

ImageFX provided by Google can generate images using Imagen 3 for free. In addition, Microsoft Bing Image Creator (equipped with DALL·E 3) and Canva's AI image generation function also have free tiers.

References