Top Rated

The highest-rated robots and AI models, ranked by the wujec.ai score.

  1. 1Claude Fable 5Anthropic · Language model · 2026Anthropic's most intelligent generally available model — the first of the Mythos-class tier above Opus.9.4no votes yet
  2. 2AlphaFold 3Google DeepMind · Science · 2024The Nobel-recognized model predicting the structure of life's molecules and their interactions.9.31 vote
  3. 3Claude Opus 5Anthropic · Language model · 2026Anthropic's reasoning-heavy Claude 5 model — one million tokens of context, adaptive thinking and an effort dial that trades cost for depth.9.3no votes yet
  4. 4Unitree As2-WUnitree Robotics · Quadruped · 2026Wheeled-legged version of the As2, shown July 2026: 25 kg, 16 legged joints, 16 kg continuous payload, ~30 km unloaded range and 6 m/s top speed.9.2no votes yet
  5. 5Claude Mythos 5Anthropic · Language model · 2026Claude Fable 5 without the safety classifiers — the same frontier model, released only to approved customers in Project Glasswing.9.2no votes yet
  6. 6da Vinci 5Intuitive Surgical · Medical · 2024The fifth generation of the world's dominant surgical robot — force feedback and 10,000× the computing power.9.2no votes yet
  7. 7Claude Opus 4.8Anthropic · Language model · 2026The last Claude 4 flagship — 41 days after Opus 4.7, with hundreds of parallel subagents, a 2.5× fast mode and a jump from 69 to 97 percent on USAMO 2026.9.2no votes yet
  8. 8GPT-4OpenAI · Language model · 2023The 2023 model that made large language models a mainstream tool — and the last one whose exam scores made headlines.9.1no votes yet
  9. 9Gemini 3.1 ProGoogle DeepMind · Language model · 2026Google's reasoning-heavy Pro model — a million tokens of context and a doubled ARC-AGI-2 score over Gemini 3 Pro.9.1no votes yet
  10. 10GPT-5.6OpenAI · Language model · 2026OpenAI's mid-2026 frontier family — the Sol, Luna and Terra line rolled out from late June.9.1no votes yet
  11. 11Claude Opus 4.7Anthropic · Language model · 2026The April 2026 Opus that pushed SWE-bench Verified past 87 percent and added an extra-high effort level above the four Opus 4.6 introduced.9.1no votes yet
  12. 12Gemini 2.5 ProGoogle DeepMind · Language model · 2025The model that made a million-token context ordinary and put thinking on by default — Google's workhorse flagship of 2025.9.1no votes yet
  13. 13GPT-5.5OpenAI · Language model · 2026OpenAI's first million-token model — an agentic-coding leader that arrived seven weeks after GPT-5.4 at double the price.9.1no votes yet
  14. 14OpenAI o3OpenAI · Language model · 2025The first model that could use tools while it was still thinking — and the one that made agentic reasoning a product rather than a demo.9.0no votes yet
  15. 15Claude 3.5 SonnetAnthropic · Language model · 2024The mid-sized model that beat its own flagship, made Claude a coding tool — and first let an AI move a mouse.9.0no votes yet
  16. 16Figure 03Figure AI · Humanoid · 2025Figure's third-generation humanoid powered by the Helix vision-language-action model; proven on BMW's production line.9.0no votes yet
  17. 17Claude Sonnet 5Anthropic · Language model · 2026Anthropic's most agentic Sonnet — 1M-token context as standard, default on free and Pro plans.9.0no votes yet
  18. 18DeepSeek-V4-ProDeepSeek · Language model · 2026DeepSeek's flagship: a 1.65-trillion-parameter MoE with open weights under MIT, whose final 0813 build lifts agentic coding to frontier level at a fraction of closed-model prices.9.0no votes yet
  19. 19Claude Opus 4.6Anthropic · Language model · 2026The first Opus with a million-token context — Anthropic's February 2026 flagship, and the model where adaptive thinking replaced the extended-thinking switch.9.0no votes yet
  20. 20Gemini 3 ProGoogle DeepMind · Language model · 2025The model that opened Google's Gemini 3 line in November 2025 — natively multimodal, a million tokens of context and, briefly, the top of the leaderboards.8.9no votes yet
  21. 21InklingThinking Machines Lab · Language model · 2026The first model from Mira Murati's Thinking Machines Lab — a 975-billion-parameter multimodal mixture of experts released under Apache 2.0 with open weights.8.9no votes yet
  22. 22Atlas (Electric)Boston Dynamics · Humanoid · 2024The fully electric successor to the legendary hydraulic Atlas — the most athletic humanoid ever built.8.9no votes yet
  23. 23OpenAI o1OpenAI · Language model · 2024The first model sold on thinking time rather than size — and the start of the reasoning era every lab now competes in.8.9no votes yet
  24. 24GLM-5.2Z.ai (Zhipu AI) · Language model · 2026The open-weight model that made frontier coding cheap — MIT licence, one-million-token context, no regional limits.8.9no votes yet
  25. 25GPT-5OpenAI · Language model · 2025The August 2025 flagship that introduced the router — one name hiding a fast model, a reasoning model and a decision about which to use. Switched off on 11 December 2026.8.9no votes yet