endorsii's picture
Added Gpt5.6, Gemini 3.6, and 3.5
38fcad7
Raw
History Blame Contribute Delete
13.5 kB
{
"schema_version": "1.0",
"benchmark_version": "1.0",
"updated": "2026-07-27",
"models": [
{
"id": "google-gemini-3-1-pro",
"name": "Gemini 3.1 Pro",
"organization": "Google",
"params": "—",
"type": "closed",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://deepmind.google/technologies/gemini/",
"overall": 61.915,
"spatial": 71.72,
"spatio_temporal": 64.35,
"temporal": 40.74,
"semantic": 70.86,
"2d_localization": 76.15,
"2d_referring_expressions": 67.17,
"2d_spatial_pointing": 71.84,
"single_object_tracking": 64.35,
"temporal_localization": 45.69,
"dense_video_captioning": 35.78,
"event_verification": 69.83,
"video_qa": 71.88
},
{
"id": "alibaba-qwen3-vl-32b",
"name": "Qwen3-VL-32B-Instruct",
"organization": "Alibaba",
"params": "32B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://huggingface.co/Qwen/Qwen3-VL-32B-Instruct",
"overall": 55.04875,
"spatial": 70.78,
"spatio_temporal": 44.19,
"temporal": 39.56,
"semantic": 65.67,
"2d_localization": 64.33,
"2d_referring_expressions": 72.39,
"2d_spatial_pointing": 75.62,
"single_object_tracking": 44.19,
"temporal_localization": 49.71,
"dense_video_captioning": 29.4,
"event_verification": 60.04,
"video_qa": 71.3
},
{
"id": "nvidia-cosmos-reason2-8b",
"name": "Cosmos-Reason2-8B",
"organization": "NVIDIA",
"params": "8B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://huggingface.co/nvidia/Cosmos-Reason2-8B",
"overall": 54.1825,
"spatial": 73.12,
"spatio_temporal": 37.69,
"temporal": 39.9,
"semantic": 66.02,
"2d_localization": 83.88,
"2d_referring_expressions": 66.88,
"2d_spatial_pointing": 68.6,
"single_object_tracking": 37.69,
"temporal_localization": 47.3,
"dense_video_captioning": 32.5,
"event_verification": 64.09,
"video_qa": 67.95
},
{
"id": "alibaba-qwen3-vl-8b",
"name": "Qwen3-VL-8B-Instruct",
"organization": "Alibaba",
"params": "8B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://huggingface.co/Qwen/Qwen3-VL-8B-Instruct",
"overall": 50.06875,
"spatial": 67.26,
"spatio_temporal": 33.12,
"temporal": 36.98,
"semantic": 62.92,
"2d_localization": 59.93,
"2d_referring_expressions": 73.29,
"2d_spatial_pointing": 68.56,
"single_object_tracking": 33.12,
"temporal_localization": 44.32,
"dense_video_captioning": 29.64,
"event_verification": 59.39,
"video_qa": 66.44
},
{
"id": "alibaba-qwen3-5-vl-27b",
"name": "Qwen3.5-27B",
"organization": "Alibaba",
"params": "27B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://huggingface.co/Qwen/Qwen3.5-VL-27B-Instruct",
"overall": 49.25333333,
"spatial": 79.04,
"spatio_temporal": 24.4,
"temporal": 31.93,
"semantic": 61.65,
"2d_localization": 85.46,
"2d_referring_expressions": 76.35,
"2d_spatial_pointing": 75.32,
"single_object_tracking": 24.4,
"temporal_localization": 36.88,
"dense_video_captioning": 26.97,
"event_verification": 55.34,
"video_qa": 67.95
},
{
"id": "alibaba-qwen3-5-vl-9b",
"name": "Qwen3.5-9B",
"organization": "Alibaba",
"params": "9B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://huggingface.co/Qwen/Qwen3.5-VL-9B-Instruct",
"overall": 48.48166667,
"spatial": 75.26,
"spatio_temporal": 23.52,
"temporal": 35.95,
"semantic": 59.21,
"2d_localization": 76.87,
"2d_referring_expressions": 77.16,
"2d_spatial_pointing": 71.74,
"single_object_tracking": 23.52,
"temporal_localization": 42.53,
"dense_video_captioning": 29.36,
"event_verification": 46.78,
"video_qa": 71.63
},
{
"id": "google-gemini-3-2-flash-lite",
"name": "Gemini 3.1 Flash Lite",
"organization": "Google",
"params": "—",
"type": "closed",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://deepmind.google/technologies/gemini/",
"overall": 51.335,
"spatial": 62.06,
"spatio_temporal": 46.68,
"temporal": 35.17,
"semantic": 61.43,
"2d_localization": 69.88,
"2d_referring_expressions": 52.62,
"2d_spatial_pointing": 63.68,
"single_object_tracking": 46.68,
"temporal_localization": 37.72,
"dense_video_captioning": 32.62,
"event_verification": 54.83,
"video_qa": 68.03
},
{
"id": "nvidia-cosmos-reason2-32b",
"name": "Cosmos-Reason2-32B",
"organization": "NVIDIA",
"params": "32B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": false,
"date_evaluated": "2026-05-27",
"model_url": "https://huggingface.co/nvidia/Cosmos-Reason2-32B",
"overall": 48.55,
"spatial": 48.88,
"spatio_temporal": 33.12,
"temporal": 40.26,
"semantic": 71.94,
"2d_localization": 22.57,
"2d_referring_expressions": 50.04,
"2d_spatial_pointing": 74.03,
"single_object_tracking": 33.12,
"temporal_localization": 50.47,
"dense_video_captioning": 30.04,
"event_verification": 73.58,
"video_qa": 70.29
},
{
"id": "nvidia-cosmos3-super",
"name": "Cosmos3-Super",
"organization": "NVIDIA",
"params": "64B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-05-29",
"model_url": "https://huggingface.co/nvidia/Cosmos3-Super",
"overall": 63.01333333,
"spatial": 78.71,
"spatio_temporal": 62.25,
"temporal": 40.72,
"semantic": 70.37,
"2d_localization": 86.97,
"2d_referring_expressions": 76.23,
"2d_spatial_pointing": 72.94,
"single_object_tracking": 62.25,
"temporal_localization": 51.9,
"dense_video_captioning": 29.54,
"event_verification": 71.28,
"video_qa": 69.46
},
{
"id": "nvidia-cosmos3-nano",
"name": "Cosmos3-Nano",
"organization": "NVIDIA",
"params": "16B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-05-29",
"model_url": "https://huggingface.co/nvidia/Cosmos3-Nano",
"overall": 60.67291667,
"spatial": 74.84,
"spatio_temporal": 59.21,
"temporal": 39.73,
"semantic": 68.92,
"2d_localization": 74.11,
"2d_referring_expressions": 75.57,
"2d_spatial_pointing": 74.83,
"single_object_tracking": 59.21,
"temporal_localization": 48.04,
"dense_video_captioning": 31.42,
"event_verification": 68.88,
"video_qa": 68.95
},
{
"id": "alibaba-qwen3-vl-2b",
"name": "Qwen3-VL-2B-Instruct",
"organization": "Alibaba",
"params": "2B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-16",
"model_url": "https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct",
"overall": 40.24583333,
"spatial": 58.28,
"spatio_temporal": 30.38,
"temporal": 18.03,
"semantic": 54.29,
"2d_localization": 56.6,
"2d_referring_expressions": 65.12,
"2d_spatial_pointing": 53.13,
"single_object_tracking": 30.38,
"temporal_localization": 35.21,
"dense_video_captioning": 0.85,
"event_verification": 44.73,
"video_qa": 63.85
},
{
"id": "nvidia-cosmos-reason2-2b",
"name": "Cosmos-Reason2-2B",
"organization": "NVIDIA",
"params": "2B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-16",
"model_url": "https://huggingface.co/nvidia/Cosmos-Reason2-2B",
"overall": 45.84958333,
"spatial": 62.83,
"spatio_temporal": 26.85,
"temporal": 33.74,
"semantic": 59.98,
"2d_localization": 72.47,
"2d_referring_expressions": 56.33,
"2d_spatial_pointing": 59.7,
"single_object_tracking": 26.85,
"temporal_localization": 38.95,
"dense_video_captioning": 28.52,
"event_verification": 55.27,
"video_qa": 64.69
},
{
"id": "nvidia-cosmos3-edge",
"name": "Cosmos3-Edge",
"organization": "NVIDIA",
"params": "4B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-16",
"model_url": "https://huggingface.co/nvidia/Cosmos3-Edge",
"overall": 46.93291667,
"spatial": 71.99,
"spatio_temporal": 24.65,
"temporal": 28.08,
"semantic": 63.02,
"2d_localization": 78.93,
"2d_referring_expressions": 69.27,
"2d_spatial_pointing": 67.76,
"single_object_tracking": 24.65,
"temporal_localization": 39.51,
"dense_video_captioning": 16.64,
"event_verification": 63.36,
"video_qa": 62.68
},
{
"id": "google-gemma-4-e2b-it",
"name": "Gemma-4-E2B-it",
"organization": "Google",
"params": "2B",
"type": "open",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-16",
"model_url": "https://huggingface.co/google/gemma-4-e2b-it",
"overall": 21.55625,
"spatial": 28.56,
"spatio_temporal": 11.58,
"temporal": 9.16,
"semantic": 36.92,
"2d_localization": 48.42,
"2d_referring_expressions": 5.52,
"2d_spatial_pointing": 31.74,
"single_object_tracking": 11.58,
"temporal_localization": 9.02,
"dense_video_captioning": 9.31,
"event_verification": 27.65,
"video_qa": 46.19
},
{
"id": "google-gemini-3-6-flash",
"name": "Gemini 3.6 Flash",
"organization": "Google",
"params": "—",
"type": "closed",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-21",
"model_url": "https://deepmind.google/technologies/gemini/",
"overall": 69.5187,
"spatial": 78.85,
"spatio_temporal": 75.99,
"temporal": 43.825,
"semantic": 79.41,
"2d_localization": 81.57,
"2d_referring_expressions": 75.78,
"2d_spatial_pointing": 79.2,
"single_object_tracking": 75.99,
"temporal_localization": 51.5,
"dense_video_captioning": 36.15,
"event_verification": 82,
"video_qa": 76.82
},
{
"id": "google-gemini-3-5-flash-lite",
"name": "Gemini 3.5 Flash-Lite",
"organization": "Google",
"params": "—",
"type": "closed",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-21",
"model_url": "https://deepmind.google/models/gemini/flash-lite/",
"overall": 52.8925,
"spatial": 61.09,
"spatio_temporal": 55.77,
"temporal": 26.7,
"semantic": 68.01,
"2d_localization": 65.73,
"2d_referring_expressions": 74.46,
"2d_spatial_pointing": 43.08,
"single_object_tracking": 55.77,
"temporal_localization": 24,
"dense_video_captioning": 29.4,
"event_verification": 70.41,
"video_qa": 65.61
},
{
"id": "openai-gpt-5-6-sol",
"name": "GPT-5.6 Sol",
"organization": "OpenAI",
"params": "—",
"type": "closed",
"result_type": "single",
"verified": true,
"ensemble": false,
"is_new": true,
"date_evaluated": "2026-07-09",
"model_url": "https://openai.com/index/gpt-5-6/",
"overall": 69.0258,
"spatial": 77.9333,
"spatio_temporal": 75.02,
"temporal": 46.25,
"semantic": 76.9,
"2d_localization": 69.3,
"2d_referring_expressions": 82.6,
"2d_spatial_pointing": 81.9,
"single_object_tracking": 75.02,
"temporal_localization": 54.8,
"dense_video_captioning": 37.7,
"event_verification": 75.9,
"video_qa": 77.9
}
]
}