Figure 01
· Changed: deployments, status
status: prototype → retired
HMX score: 70.9 → 69.0
Evidence snapshots 210 (before) and 493 (after): through the API
Watch catalogue changes
Choose up to 20 robots — by name or by slug, the last part of a robot page's address — and change categories. Bookmark this selection or subscribe to its RSS feed in your feed reader. For security, the key in each RSS link rotates every week, so come back here weekly for a fresh link. Removing the feed in your reader stops the subscription. This site does not send email notifications.
No category selection means all change types.
Subscribe via RSS · Export changes and cited research as JSON (API access)
The RSS key rotates every week: this feed link works until 6 Nov 2026, 00:00 UTC. After that your feed reader shows a notice linking back here, where the new link is waiting — replace the old one with it.
Changes since snapshot tracking began come from recorded snapshots; earlier ones come from approved updates, with before and after values as recorded there. Shows at most 100 events from the latest 101 snapshots per robot and latest 1,000 selected-robot approvals. Older versions remain available in each robot's paginated history. A catalogue change is not automatically a verified improvement.
- Figure 01 — 12 changes shown below, records since 23 Aug 2026.
Figure 01
· Approved update · Changed: status, ai model
status: retired → prototype
ai model: OpenAI-trained large multimodal model (vision + language) for high-level reasoning; Figure neural network for visuomotor control (pixels to actions at 10 Hz input, 200 Hz output) → OpenAI large multimodal model (vision + language) for perception, reasoning, and dialogue; Figure's own neural-network visuo-motor policies for real-time action
Figure 01
· Approved update · Changed: ai model, onboard compute, height cm
ai model: OpenAI large multimodal model → OpenAI-trained large multimodal model (vision + language) for high-level reasoning; Figure neural network for visuomotor control (pixels to actions at 10 Hz input, 200 Hz output)
onboard compute: GPU → GPU (specific model not disclosed)
height cm: 167.64 → 168
Figure 01
· Approved update · Changed: ai model
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI large multimodal model
Figure 01
· Approved update · Changed: ai model
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI-trained large multimodal model for perception, dialogue, planning, and behaviour selection; Figure neural-network visuomotor policies (pixel-to-action, 10 Hz image input → 200 Hz action output) for dexterous motion
Figure 01
· Approved update · Changed: ai model
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI-trained large multimodal model (vision + language) for high-level reasoning; Figure neural-network visuomotor policies mapping pixels to actions (10 Hz image input, 200 Hz action output)
Figure 01
· Approved update · Changed: ai model, onboard compute
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI Large Multimodal Model + Figure Visuomotor Neural Network Policy
onboard compute: — → GPU
Figure 01
· Approved update · Changed: ai model, onboard compute, first unveiled year
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI large multimodal model (vision + language) for cognition; Figure's own neural-network visuomotor policies for action
onboard compute: — → GPU (specific chipset not stated)
first unveiled year: 2023 → 2024
Figure 01
· Approved update · Changed: status, ai model, onboard compute, hand dof per hand, deployments
status: prototype → retired
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI VLM/LLM (terminated Feb 2025)
onboard compute: — → Custom onboard compute
hand dof per hand: — → 6
Figure 01
· Approved update · Changed: autonomy level, ai model, height cm
autonomy level: supervised-autonomy → task-autonomous
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI multimodal model
height cm: 168 → 167.64
Figure 01
· Approved update · Changed: ai model, onboard compute
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI multimodal model + Figure visuomotor neural network policy
onboard compute: — → Onboard GPU
Figure 01
· Approved update · Changed: ai model
ai model: OpenAI vision-language model (speech-to-speech integration) → OpenAI multimodal large model for visual reasoning and language understanding, paired with Figure neural networks for visuomotor control
What Watch tracks
Watch follows up to 20 of the 222 robots in the index and reports what changes in them — evidence, deployments, specs, price, availability and tests — as a page, an RSS feed or JSON. It records that a value changed and where the new value came from; it does not judge whether the change is an improvement.
Recently changed across the catalogue: Optimus Gen 3, Surveyor, Kangaroo, SPACEO M1, SA01. Watch these.