List your company

Long-horizon game control on StarDojo Lite-100

20.3Lite-100 Success Rate (SR)SPIKE
6.1569.82813.517.172May 18, 2026

Evaluation results

MethodLinks
SPIKELLM Backend=GPT-5.42026.0520.323.7107.827.93.5
SPIKELLM Backend=Default Qwen3.5-397B2026.051821.6168.143.52.75
SPIKELLM Backend=Gemini-3.1-pro2026.0517.720.3134.333.82.66
CRADLELLM Backend=Qwen3.5-397B-A17B2026.051310.7372.573.52.49
StarDojoLLM Backend=Qwen3.5-397B-A17B2026.0512.312.3295.948.72.37
Reflexion-likeLLM Backend=Qwen3.5-397B-A17B2026.058.710.4178.535.92.02
Voyager-likeLLM Backend=Qwen3.5-397B-A17B2026.057.38.1238.356.51.73
ReAct-likeLLM Backend=Qwen3.5-397B-A17B2026.056.79.6102.624.51.61