{
  "verifiedDate": "2026-09-16",
  "models": [
    {
      "id": "llama-3-1-8b-instruct",
      "name": "Llama 3.1 8B Instruct",
      "summary": "Text assistant supporting eight languages, with a 128k context window and instruction tuning for dialogue.",
      "releaseDate": "2024-07-23",
      "organization": "Meta",
      "availability": "Open weights",
      "license": {
        "label": "Llama 3.1 Community License Agreement",
        "url": "https://github.com/meta-llama/llama-models/blob/main/models/llama3_1/LICENSE"
      },
      "access": "Downloadable weights; custom license and acceptable-use restrictions; gated access on Hugging Face.",
      "type": "LLM (text → text)",
      "architecture": "Autoregressive Transformer with grouped-query attention; SFT and RLHF instruction tuning.",
      "parameters": "8 billion (reported model size)",
      "benchmarks": [
        {
          "name": "MMLU",
          "score": "69.4%",
          "setup": "Instruction-tuned 8B; 5-shot; macro_avg/acc; Meta internal evaluation library. Dataset revision and library version: Not reported.",
          "source": {
            "label": "Meta model card",
            "url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct"
          },
          "sortScore": 69.4,
          "sortMetric": "Accuracy (%)"
        },
        {
          "name": "HumanEval",
          "score": "72.6%",
          "setup": "Instruction-tuned 8B; 0-shot; pass@1; Meta internal evaluation library. Dataset revision and sampling settings: Not reported.",
          "source": {
            "label": "Meta model card",
            "url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct"
          },
          "sortScore": 72.6,
          "sortMetric": "Pass@1 (%)"
        }
      ],
      "paper": {
        "label": "The Llama 3 Herd of Models",
        "url": "https://arxiv.org/abs/2407.21783"
      },
      "code": {
        "label": "Meta reference implementation",
        "url": "https://github.com/meta-llama/llama-models"
      },
      "sources": [
        {
          "label": "Meta model card",
          "url": "https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct"
        }
      ],
      "architectureFamily": "Autoregressive Transformer",
      "organizationIconDomain": "meta.com",
      "organizationSymbol": "M"
    },
    {
      "id": "gpt-4-original",
      "name": "GPT-4 (original, 2023)",
      "summary": "Multimodal model for image and text understanding and text generation. This entry describes the original report, rather than later GPT-4 variants.",
      "releaseDate": "2023-03-14",
      "organization": "OpenAI",
      "availability": "Proprietary",
      "license": {
        "label": "No public model license; hosted service governed by OpenAI terms",
        "url": "https://openai.com/policies/terms-of-use/"
      },
      "access": "No released weights. At launch: text access through ChatGPT and an API waitlist; image access limited. Historical release, not a current availability guarantee.",
      "type": "Multimodal LLM / VLM (image + text → text)",
      "architecture": "Transformer trained for next-token prediction and aligned with RLHF; detailed architecture: Not reported.",
      "parameters": null,
      "benchmarks": [
        {
          "name": "MMLU",
          "score": "86.4%",
          "setup": "Original 2023 technical report, Table 2; 5-shot accuracy over 57 subjects; text evaluation. Dataset revision: Not reported. Report notes minor differences from standard evaluation setups.",
          "source": {
            "label": "Technical report, Table 2",
            "url": "https://cdn.openai.com/papers/gpt-4.pdf"
          },
          "sortScore": 86.4,
          "sortMetric": "Accuracy (%)"
        }
      ],
      "paper": {
        "label": "Technical report, Table 2",
        "url": "https://cdn.openai.com/papers/gpt-4.pdf"
      },
      "code": null,
      "sources": [
        {
          "label": "Launch announcement",
          "url": "https://openai.com/index/gpt-4-research/"
        },
        {
          "label": "Technical report, Table 2",
          "url": "https://cdn.openai.com/papers/gpt-4.pdf"
        }
      ],
      "architectureFamily": "Transformer (details not reported)",
      "organizationIconDomain": "openai.com",
      "organizationSymbol": "O"
    },
    {
      "id": "clip-vit-b-32",
      "name": "CLIP ViT-B/32",
      "summary": "Contrastively trained image and text encoders for retrieval and zero-shot classification using natural-language class descriptions.",
      "releaseDate": "2021-01-05",
      "organization": "OpenAI",
      "availability": "Open source",
      "license": {
        "label": "MIT License (released code and model weights)",
        "url": "https://github.com/openai/CLIP/blob/main/LICENSE"
      },
      "access": "Public code and pretrained weights; training dataset is not released.",
      "type": "VLM (image–text embedding model)",
      "architecture": "Dual encoder: ViT-B/32 image encoder and Transformer text encoder; contrastive image–text training.",
      "parameters": null,
      "benchmarks": [
        {
          "name": "ImageNet-1K",
          "score": "63.2% top-1",
          "setup": "CLIP paper Table 17; ViT-B/32, 224px input; zero-shot classification with prompt ensembling, no ImageNet classifier training. Evaluation uses ImageNet validation; dataset revision: Not reported.",
          "source": {
            "label": "CLIP paper, Table 17",
            "url": "https://cdn.openai.com/papers/Learning_Transferable_Visual_Models_From_Natural_Language_Supervision.pdf"
          },
          "sortScore": 63.2,
          "sortMetric": "Top-1 accuracy (%)"
        }
      ],
      "paper": {
        "label": "Learning Transferable Visual Models From Natural Language Supervision",
        "url": "https://arxiv.org/abs/2103.00020"
      },
      "code": {
        "label": "OpenAI CLIP",
        "url": "https://github.com/openai/CLIP"
      },
      "sources": [
        {
          "label": "Release announcement",
          "url": "https://openai.com/index/clip/"
        },
        {
          "label": "CLIP paper, Table 17",
          "url": "https://cdn.openai.com/papers/Learning_Transferable_Visual_Models_From_Natural_Language_Supervision.pdf"
        },
        {
          "label": "Official model card",
          "url": "https://github.com/openai/CLIP/blob/main/model-card.md"
        }
      ],
      "architectureFamily": "Dual encoder (ViT + text Transformer)",
      "organizationIconDomain": "openai.com",
      "organizationSymbol": "O"
    },
    {
      "id": "deit-base-224",
      "name": "DeiT-Base (224, non-distilled)",
      "summary": "Image classifier trained on ImageNet without external training data. This entry is the baseline model; the paper also introduces attention-based distillation variants.",
      "releaseDate": "Not reported (paper first submitted 2020-12-23)",
      "organization": "Meta FAIR",
      "availability": "Open source",
      "license": {
        "label": "Apache License 2.0 (repository release)",
        "url": "https://github.com/facebookresearch/deit/blob/main/LICENSE"
      },
      "access": "Public code and baseline pretrained checkpoints linked in the official repository.",
      "type": "Vision model (image classification)",
      "architecture": "Vision Transformer, base size; 16×16 image patches; 224×224 input; non-distilled baseline.",
      "parameters": "86 million (official model zoo)",
      "benchmarks": [
        {
          "name": "ImageNet-1K",
          "score": "81.8% top-1; 95.6% top-5",
          "setup": "Official baseline DeiT-base checkpoint; ImageNet 2012 training only, no external data; 224px, single-crop validation. Reference inference uses timm 0.3.2. Distilled and 384px variants have different scores.",
          "source": {
            "label": "Official model zoo and evaluation instructions",
            "url": "https://github.com/facebookresearch/deit/blob/main/README_deit.md"
          },
          "sortScore": 81.8,
          "sortMetric": "Top-1 accuracy (%)"
        }
      ],
      "paper": {
        "label": "Training data-efficient image transformers & distillation through attention",
        "url": "https://arxiv.org/abs/2012.12877"
      },
      "code": {
        "label": "Official DeiT repository",
        "url": "https://github.com/facebookresearch/deit"
      },
      "sources": [
        {
          "label": "Official model zoo and evaluation instructions",
          "url": "https://github.com/facebookresearch/deit/blob/main/README_deit.md"
        },
        {
          "label": "Paper and author affiliations",
          "url": "https://arxiv.org/abs/2012.12877"
        }
      ],
      "architectureFamily": "Vision Transformer",
      "organizationIconDomain": "meta.com",
      "organizationSymbol": "M"
    }
  ]
}