Can I use many AI models without separate subscriptions?
Yes. FLORA puts image, video, audio, text and 3D models from many providers behind one account, billed per run from one workspace. In our run, one prompt went to five image models from five providers in a single call, and all five images came back in under 36 seconds for $0.34 in total, with no other sign-ups.
Key takeaways
- FLORA's model list showed 237 models on October 16, 2026, across image, video, audio, text and 3D.
- One call sent the same prompt to GPT Image 2.5 Flare, Nano Banana 2.1, Krea 3, Ideogram 4.5 and FLUX 3 Image.
- The five images cost between $0.05 and $0.12 each, $0.34 in total.
- Every run is billed in dollars to one FLORA workspace, and any run can be quoted first.
At a glance
| What it does | Runs models from many providers from one account and one bill |
|---|---|
| Works in | The FLORA canvas, or Claude, ChatGPT, Cursor and Codex through the FLORA MCP |
| Models | 237 listed, including GPT Image 2.5, Nano Banana, Krea, Ideogram, FLUX, Seedance, Lyria and Rodin |
| Inputs | A prompt, plus images, video or 3D for models that take them |
| Outputs | Images, video, audio, text and 3D models, saved to one canvas |
| Typical cost | $0.05 to $0.12 per image in our comparison |
| Typical time | 18 to 35 seconds per image, all in parallel |
Key features
- One bill: Every run is charged in dollars to your FLORA workspace, whichever provider made it.
- Quotes before runs: Ask for a quote and FLORA returns the price of any model run with nothing reserved.
- Side-by-side batches: Up to 20 generations, on any mix of models, go out in one call.
- Shared canvas: Outputs from every model land on the same canvas, so one model's image can feed another's input.
Best for
- Teams comparing models before picking one for a project
- Workflows that chain models: image, then video, then music
- Freelancers who do not want five subscriptions for occasional use
- Agents that pick the right model per task
Step by step
Step 1: Connect once
Sign in to FLORA on the canvas, or add the FLORA MCP to your agent. That one sign-in covers every model.
Step 2: List the models
Ask the agent to list models by type. It returns ids, capabilities and an estimated cost for each.
Step 3: Run one prompt on several models
Send the same prompt to each model in one batch, then compare the results on the canvas.
Run this prompt through FLORA on GPT Image 2.5 Flare, Nano Banana 2.1, Krea 3, Ideogram 4.5 and FLUX 3 Image in one batch, then show me all five with their cost: A glass jar of wildflower honey with a wooden dipper on a kitchen counter, morning window light, a slice of sourdough toast beside it. Natural muted colors, true-to-life photograph.Step 4: Keep the winner, chain the next step
Pick the best image and pass it to the next model: an upscaler, a video model or an image-to-3D model.
What happened when we ran several models on one FLORA account?
On October 16, 2026, the agent called flora_list_models, which reported 237 models. It then sent one flora_create_generations call with five items: the honey prompt from step 3, each item naming a different model id and no other settings. One FLORA sign-in covered all five.
All five finished within 18 to 35 seconds of submission. GPT Image 2.5 Flare (t2i-gpt-image-2-5-flare) took 35 seconds and cost $0.055. Nano Banana 2.1 (t2i-elote-gateway) took 25 seconds and cost $0.052. Krea 3 (t2i-krea-3-t2i) took 18 seconds and cost $0.063. Ideogram 4.5 (t2i-ideogram-4-5-t2i) took 26 seconds and cost $0.12. FLUX 3 Image (t2i-flux-3-image-t2i) took 35 seconds and cost $0.05. Total: $0.34.
All five returned the jar, the dipper, the toast and the window in soft light. They differed in the details the prompt left open. GPT Image 2.5 and Nano Banana 2.1 added a printed "Wildflower Honey" label, Ideogram 4.5 printed WILDFLOWER on a plain label, and Krea 3 and FLUX 3 Image left the jar unlabelled. Ideogram 4.5 and FLUX 3 Image showed honey running from the dipper.
What went wrong: with no size set, the outputs came back at different resolutions, 2048 x 2048 from Krea 3 and Ideogram 4.5 and 1024 x 1024 from the other three. That makes a fair side-by-side harder. The fix is to set aspect ratio and resolution per model, using the parameter names flora_list_models returns for each.
- Models
- 5 image models from 5 providers
- Cost
- $0.34 in total, $0.05 to $0.12 each
- Time
- 18 to 35 s, parallel
- Accounts
- 1 FLORA sign-in
| Model | Model id | Cost | Time | Size |
|---|---|---|---|---|
| GPT Image 2.5 Flare | t2i-gpt-image-2-5-flare | $0.055 | 35 s | 1024 x 1024 |
| Nano Banana 2.1 | t2i-elote-gateway | $0.052 | 25 s | 1024 x 1024 |
| Krea 3 | t2i-krea-3-t2i | $0.063 | 18 s | 2048 x 2048 |
| Ideogram 4.5 | t2i-ideogram-4-5-t2i | $0.12 | 26 s | 2048 x 2048 |
| FLUX 3 Image | t2i-flux-3-image-t2i | $0.05 | 35 s | 1024 x 1024 |
One model or a side-by-side batch: pros and cons
Pros
- One model: cheapest per idea once you know which model fits.
- Side-by-side batch: shows which model handles your subject best before you commit.
- Side-by-side batch: all results arrive together, in parallel.
Cons
- One model: you never see what another model would have done.
- Side-by-side batch: costs one run per model.
- Side-by-side batch: output sizes differ by model unless you set them.
How does billing work across providers?
Each run is charged in dollars to your FLORA workspace at that model's price. You do not open an account with the model's provider. Ask the agent to quote a run, or a whole batch, before it starts.
Which kinds of models are included?
The list covers every step of a visual project.
- Image: GPT Image 2.5, Nano Banana, Krea, Ideogram, FLUX, Seedream, Recraft and more.
- Video: Seedance 2.5 and other image-to-video and text-to-video models.
- Audio: Lyria 3.5, Suno v6, ElevenLabs music, sound effects and voices.
- 3D: Rodin, Meshy, Tripo, Hunyuan3D and Trellis image-to-3D and text-to-3D.
- Upscalers and text models for captions and image descriptions.
Can one model's output feed another?
Yes. Everything lands on the same FLORA canvas, so an image from one model can be the input to a video model, an upscaler or an image-to-3D model without downloading and re-uploading.
Frequently asked questions
Do I need an account with each AI provider?
No. One FLORA account runs every model on the list, and each run is billed to your workspace.
How many models can I use?
FLORA's model list showed 237 models on October 16, 2026.
How do I compare models?
Send one prompt to several models in a single batch. Our five-model comparison cost $0.34 and finished in 35 seconds.
Can my agent pick the model?
Yes. Name a model, or let the agent choose from the list based on the task. It can quote the cost before it runs.