+1 (415) 347-6981Get started
Fixed prices · You own the code · Proposal in 24h hello@startrise.io →
3D / game

Playable 3D Game

30 models built this brief single-shot. Claude Opus 5 leads at 88.0. 30 of 30 builds carry a blind human rating. 2 split the judge panel.

GitHub repo ↗
↗
  1. Claude Opus 5 Anthropic · Jul 2026
    88.0
    100 gates80 panel90 human
    Open live ↗
  2. Claude Sonnet 5.5 Anthropic · Oct 2026
    86.5
    100 gates80 panel85 human
    Open live ↗
  3. Claude Sonnet 5.5 (high) Anthropic · Oct 2026
    85.0
    100 gates80 panel80 human
    Open live ↗
  4. GPT-6 Astra OpenAI · Oct 2026
    83.1
    100 gates83 panel70 human
    Open live ↗
  5. Claude Opus 5.5 Anthropic · Oct 2026
    80.1
    100 gates83 panel60 human
    Open live ↗
  6. Muse Spark 1.3 Meta · Oct 2026
    79.8
    100 gates75 panel70 human
    Open live ↗
  7. Kimi K3 Moonshot · Jul 2026
    78.6
    100 gates73 panel70 human
    Open live ↗
  8. DeepSeek V4.1 Flash DeepSeek · Oct 2026
    76.9
    98 gates70 panel70 human
    Open live ↗
  9. Claude Fable 5 Anthropic · Jul 2026
    76.8
    100 gates65 panel75 human
    Open live ↗
  10. GPT-6.1 Sol OpenAI · Oct 2026
    74.9
    100 gates78 panel50 human
    Open live ↗
  11. Claude Opus 5.5 (high) Anthropic · Oct 2026
    74.9
    100 gates78 panel50 human
    Open live ↗
  12. Grok 4.5 xAI · Jul 2026
    73.2
    98 gates65 panel65 human
    Open live ↗
  13. GPT-5.6 Luna OpenAI · Jul 2026
    72.3
    100 gates65 panel60 human
    Open live ↗
  14. Claude Fable 5.1 Anthropic · Oct 2026
    70.4
    100 gates68 panel50 human
    Open live ↗
  15. GPT-5.6 Terra OpenAI · Jul 2026
    69.3
    100 gates65 panel50 human
    Open live ↗
  16. Qwen 3.8 Max 0902 Alibaba · Oct 2026
    68.6
    91 gates75 panel40 human
    Open live ↗
  17. GLM 5.3 Z.ai · Oct 2026
    68.1
    100 gates63 panel50 human
    Open live ↗
  18. Muse Spark 1.2 Meta · Aug 2026
    67.6
    92 gates63 panel55 human
    Open live ↗
  19. Gemini 3.8 Flash Google · Oct 2026 · split panel
    64.9
    96 gates58 panel50 human
    Open live ↗
  20. GLM 5.2 Z.ai · Jul 2026
    64.4
    100 gates58 panel45 human
    Open live ↗
  21. Claude Sonnet 5 Anthropic · Jul 2026
    63.6
    100 gates63 panel35 human
    Open live ↗
  22. Grok 4.7 xAI · Oct 2026
    60.4
    99 gates73 panel10 human
    Open live ↗
  23. DeepSeek V4 Pro 0813 DeepSeek · Oct 2026
    56.2
    91 gates58 panel25 human
    Open live ↗
  24. Qwen 3.7 Max Alibaba · Jul 2026
    55.0
    100 gates50 panel25 human
    Open live ↗
  25. DeepSeek V4 Pro DeepSeek · Jul 2026
    54.5
    94 gates53 panel25 human
    Open live ↗
  26. Qwen 3.8 Max Alibaba · Aug 2026
    47.7
    86 gates45 panel20 human
    Open live ↗
  27. GPT-5.6 Sol OpenAI · Jul 2026 · split panel
    46.1
    64 gates50 panel25 human
    Open live ↗
  28. Claude Haiku 4.5 Anthropic · Jul 2026
    43.2
    86 gates35 panel20 human
    Open live ↗
  29. GPT-6 Luna OpenAI · Oct 2026
    28.5
    64 gates28 panel0 human
    Open live ↗
  30. Grok 4.6 xAI · Aug 2026
    26.2
    64 gates23 panel0 human
    Open live ↗

Scores are the blended final: 25% browser gates, 45% blind three-lab panel, 30% blind human rating; unrated cells redistribute the human weight. Screenshots are the gate run at 1440×900. "Run it live" loads the untouched deliverable in a sandboxed frame - bugs included. Back to the leaderboard · Compare the top two