Model comparison
new_reward_dataset_irrh6c0s vs GPT-5
Capability radar, value bars, and side-by-side posture.
new_reward_dataset_irrh6c0s
Not scored
value score
GPT-5
94
value score
new_reward_dataset_irrh6c0s ctx
—
GPT-5 ctx
400K
Capability radar
Value · context · multimodal · openness · speed posture
Value head-to-head
- GPT-594
Attribute tape
| Field | new_reward_dataset_irrh6c0s | GPT-5 |
|---|---|---|
| Developer | strawberryroboticsinc | OpenAI |
| Context | — | 400K |
| Modalities | Text, Code, Image, Audio | |
| Openness | — | Proprietary Model |
| Speed | — | Fast |
| Price | — | Mid-High |
| Value score | — | 94 |
| Summary | A proprietary robotics reward dataset published by Strawberry Robotics, Inc. on Hugging Face, designed to support reward-model training and evaluation for robot task performance. The dataset contains trajectory, video, or observation data paired with reward labels, following common embodied-AI reward-modeling pipelines. | OpenAI frontier multimodal model. |