Leaderboard
Leaderboard is A Unitree G1 benchmark of 20 household tasks with a frozen whole-body controller in the loop, 60 human VR demonstrations per task, and a coding-agent track scored on the same hidden seeds as learned policies.. It is ranked #689 on the AI Radar, in AI research, first seen 11 days ago and shared in 2 posts (20.6k views).
Leaderboard · BiGym 2.0 Coding agents Each agent writes the controller for a whole-body humanoid from one demonstration video, with the same 101,000 environment steps per session. Mean success over 9 tasks × 3 independent sessions, 100 hidden episodes per final program. Success against compute Agent Success Cost per task Output tokens per task GPT-6 Astra high · Codex 64.5% $11.0 34k Claude Opus 5.5 high · Claude Code 60.6% $15.6 192k GPT-6.1 Sol high · Codex 40.7% $1.3 41k GPT-6 Sol high · Codex 31.2% $4.9 61k Kimi K3.1 Pro max · Claude Code 21.3% $12.1 126k DeepSeek V4.1 Flash (1017) max · dsh 13.1% $0.6 254k Every agent had the same…
What people said about Leaderboard on X
people are not noticing enough that on many nontrivial long tail capabilities, both OpenAI and Anthropic are now showing a pattern where their models reach the same or *higher* perf on lower tier models released later (Fable 5.1-5.5 Opus-Sonnet, 6 Astra - 6.1 Sol - 6.1 Terra)
— @teortaxesTex, 3 days ago · 214 likes · see the post
Now that's more like it. K3.1 is between 6 Sol and 1017. Kimi was earlier to high quality vision, it's much bigger, it should perform well on robotic/spatial/agent tasks… the issue is, again, that it costs 20 times more. Sol 6.1 must be the north star of Chinese models now.
— @teortaxesTex, 11 days ago · 100 likes · see the post
Alternatives to Leaderboard
- academy.dair.ai — Chat with Paper
- alphaXiv — Now on @askalphaxiv
- flashprove — Further progress in provable inference will likely be an important part of improving trust in AI deployment. To that…
- moe-blogpost — A hands-on guide to training Mixture-of-Experts models
- liberai.com — Technical blog
- Science's Last Exam — Can AI do research, end to end? Science's Last Exam defines and measures whether an AI agent can carry a research goal…
Leaderboard in numbers
- Rank on the AI Radar: #689 of 10402
- Shared in 2 posts by 1 account: @teortaxesTex
- 20.6k views on those posts
- First seen 11 days ago, last shared 3 days ago
- Pricing seen by AGK Intelligence: not stated
- Market: AI research
FAQ
What is Leaderboard?
A Unitree G1 benchmark of 20 household tasks with a frozen whole-body controller in the loop, 60 human VR demonstrations per task, and a coding-agent track scored on the same hidden seeds as learned policies. It was first shared on X 11 days ago and is ranked #689 on the AI Radar.
Is Leaderboard free?
Pricing is not stated on the page we read.
Who shared Leaderboard?
1 account on X, including @teortaxesTex, in 2 posts totalling 20.6k views.
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.6k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 15:10 UTC. Full method.