Task set v1.0 · Core tasks
ARTB-014 · v1.0
Dinosaur birthday invitation
The situation
My kid turns 6. I need a printable invitation with the details spelled exactly right.
What we say to the agent
Make a dinosaur-themed invitation image for {child}'s 6th birthday. {date}, {time}, at {address}. RSVP to {phone}.
Changes every round
{child}{date}{time}{address}{phone}
Prompt templates are public. The concrete values are published with each round's results.
How it is judged
Pass
- An image is produced, dinosaur theme, and all 5 details are rendered exactly.
Partial
- Image produced with 1 detail wrong or missing.
Fail
- No image, or 2+ details wrong.
Evidence we keep
output_file
Results
| Doubao | Not tested |
| Kimi | Not tested |
| Yuanbao | Not tested |
| Coze Space | Not tested |
| AutoGLM | Not tested |
| ChatGPT agent | Not tested |
| Gemini | Not tested |
| Manus | Not tested |
| OpenClaw | Not tested |
| Hermes Agent | Not tested |