{
  "schema_version": 1,
  "env": "osworld",
  "commit_dir": "2026-07-18T17-00_65ea6496ef",
  "run_id": "run_0",
  "commit": {
    "short_sha": "65ea6496ef",
    "subject": "update sys prompt for gpt"
  },
  "artifact_root": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0",
  "result_count": 13,
  "status_counts": {
    "complete": 12,
    "partial": 1,
    "empty": 0
  },
  "results": [
    {
      "model": "gpt-5.5",
      "slug": "gpt-5.5",
      "base_slug": "gpt-5.5",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.7233901272056157,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/gpt-5.5/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/gpt-5.5"
    },
    {
      "model": "Qwen/Qwen3.5-27B",
      "slug": "Qwen_Qwen3.5-27B",
      "base_slug": "Qwen_Qwen3.5-27B",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.4618772595770448,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-27B/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-27B"
    },
    {
      "model": "Qwen/Qwen3.5-9B",
      "slug": "Qwen_Qwen3.5-9B",
      "base_slug": "Qwen_Qwen3.5-9B",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.380677583188961,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-9B/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-9B"
    },
    {
      "model": "Qwen/Qwen3-VL-32B-Instruct",
      "slug": "Qwen_Qwen3-VL-32B-Instruct",
      "base_slug": "Qwen_Qwen3-VL-32B-Instruct",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.3583037643298162,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-32B-Instruct/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-32B-Instruct"
    },
    {
      "model": "Qwen/Qwen3-VL-8B-Instruct",
      "slug": "Qwen_Qwen3-VL-8B-Instruct",
      "base_slug": "Qwen_Qwen3-VL-8B-Instruct",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.34243534790271335,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-8B-Instruct/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-8B-Instruct"
    },
    {
      "model": "Qwen/Qwen3.5-4B",
      "slug": "Qwen_Qwen3.5-4B",
      "base_slug": "Qwen_Qwen3.5-4B",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.30332181956208365,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-4B/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-4B"
    },
    {
      "model": "ByteDance-Seed/UI-TARS-1.5-7B",
      "slug": "ByteDance-Seed_UI-TARS-1.5-7B",
      "base_slug": "ByteDance-Seed_UI-TARS-1.5-7B",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.29429870797822194,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/ByteDance-Seed_UI-TARS-1.5-7B/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/ByteDance-Seed_UI-TARS-1.5-7B"
    },
    {
      "model": "Qwen/Qwen3-VL-4B-Instruct",
      "slug": "Qwen_Qwen3-VL-4B-Instruct",
      "base_slug": "Qwen_Qwen3-VL-4B-Instruct",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.29070852880824405,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-4B-Instruct/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-4B-Instruct"
    },
    {
      "model": "Qwen/Qwen3-VL-2B-Instruct",
      "slug": "Qwen_Qwen3-VL-2B-Instruct",
      "base_slug": "Qwen_Qwen3-VL-2B-Instruct",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.2048280864292601,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-2B-Instruct/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3-VL-2B-Instruct"
    },
    {
      "model": "OpenGVLab/ScaleCUA-7B",
      "slug": "OpenGVLab_ScaleCUA-7B",
      "base_slug": "OpenGVLab_ScaleCUA-7B",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.15031448048319754,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/OpenGVLab_ScaleCUA-7B/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/OpenGVLab_ScaleCUA-7B"
    },
    {
      "model": "ByteDance-Seed/UI-TARS-7B-DPO",
      "slug": "ByteDance-Seed_UI-TARS-7B-DPO",
      "base_slug": "ByteDance-Seed_UI-TARS-7B-DPO",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.14742027616297482,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/ByteDance-Seed_UI-TARS-7B-DPO/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/ByteDance-Seed_UI-TARS-7B-DPO"
    },
    {
      "model": "Qwen/Qwen3.5-2B",
      "slug": "Qwen_Qwen3.5-2B",
      "base_slug": "Qwen_Qwen3.5-2B",
      "config_id": "default",
      "status": "complete",
      "num_valid": 325,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.10712923160405287,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-2B/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/Qwen_Qwen3.5-2B"
    },
    {
      "model": "meituan/EvoCUA-8B-20260105",
      "slug": "meituan_EvoCUA-8B-20260105",
      "base_slug": "meituan_EvoCUA-8B-20260105",
      "config_id": "default",
      "status": "partial",
      "num_valid": 324,
      "num_tasks": 325,
      "num_samples": 325,
      "mean_episode_return": 0.42814823921095263,
      "summary_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/meituan_EvoCUA-8B-20260105/summary.json",
      "artifact_path": ".exps/eval/osworld/2026-07-18T17-00_65ea6496ef/run_0/meituan_EvoCUA-8B-20260105"
    }
  ]
}
