{
  "_comment": "Consolidated model information mapping. Contains developer, type, cost information, and execution specifications for each model.",
  "_units": {
    "api_costs": "cost per 1M input tokens + cost per 1M output tokens (if applicable)",
    "local_costs": "cost per 1M tokens (calculated using formula)"
  },
  "anthropic/claude-haiku-4-5": {
    "model_developer": "Anthropic",
    "model_type": "generalist",
    "url": "https://platform.claude.com/docs/en/about-claude/models/overview",
    "cost_info": {
      "cost_per_1M_input_tokens": 1.0,
      "cost_per_1M_output_tokens": 5.0,
      "source": "Claude API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "anthropic/claude-sonnet-4-5": {
    "model_developer": "Anthropic",
    "model_type": "generalist",
    "url": "https://platform.claude.com/docs/en/about-claude/models/overview",
    "cost_info": {
      "cost_per_1M_input_tokens": 3.0,
      "cost_per_1M_output_tokens": 15.0,
      "source": "Claude API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/claude-latency-rerun": {
    "model_developer": "Anthropic",
    "model_type": "generalist",
    "url": "https://platform.claude.com/docs/en/about-claude/models/overview",
    "cost_info": {
      "cost_per_1M_input_tokens": 3.0,
      "cost_per_1M_output_tokens": 15.0,
      "source": "Claude API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "anthropic/claude-sonnet-4-6": {
    "model_developer": "Anthropic",
    "model_type": "generalist",
    "url": "https://platform.claude.com/docs/en/about-claude/models/overview",
    "cost_info": {
      "cost_per_1M_input_tokens": 3.0,
      "cost_per_1M_output_tokens": 15.0,
      "source": "Claude API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "azure/analyze-text": {
    "model_developer": "Azure",
    "model_type": "specialized",
    "url": "https://learn.microsoft.com/en-us/azure/ai-services/content-safety/overview",
    "cost_info": {
      "cost_per_1M_input_tokens": 380.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "Azure AI Content Safety",
      "cost_per_h": "N/A",
      "additional_info": "Usage is measured in text records. One text record may contain up to 1000 characters. Only the input string is considered in the text record.\n1000 text records cost $0.38. Assuming an average length of 500 characters for a message, and a token length of approx. 4 characters: Estimated cost per 1M input tokens = 1M/(500/4) text records * $0.38/1000 text records) = $3.04.\nCost per 1M input tokens (not records) = 3.04$.\n\nThe benchmark was run on the free plan."
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "azure/promptshield": {
    "model_developer": "Azure",
    "model_type": "specialized",
    "url": "https://learn.microsoft.com/en-us/azure/ai-foundry/openai/concepts/content-filter-prompt-shields?view=foundry-classic#prompt-shields-for-user-prompts",
    "cost_info": {
      "cost_per_1M_input_tokens": 380.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "Azure AI Content Safety",
      "cost_per_h": "N/A",
      "additional_info": "Usage is measured in text records. One text record may contain up to 1000 characters. Only the input string is considered in the text record.\n1000 text records cost $0.38. Assuming an average length of 500 characters for a message, and a token length of approx. 4 characters: Estimated cost per 1M input tokens = 1M/(500/4) text records * $0.38/1000 text records) = $3.04.\nCost per 1M input tokens (not records) = 3.04$.\n\nThe benchmark was run on the free plan."
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "google/gemini-2.5-flash": {
    "model_developer": "Google",
    "model_type": "generalist",
    "url": "https://ai.google.dev/gemini-api/docs/models#gemini-2.5-flash",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.3,
      "cost_per_1M_output_tokens": 2.5,
      "source": "Gemini Developer API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "lakera/lakera-guard_default": {
    "model_developer": "Lakera",
    "model_type": "specialized",
    "url": "https://www.lakera.ai/lakera-guard",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "Lakera API pricing",
      "cost_per_h": "N/A",
      "additional_info": "Free for 10,000 API requests/month; Allows prompt size up to 8,000 tokens per request."
    },
    "execution_specifications": {
      "type": "API",
      "details": "This is the LakeraGuard API with the default policy. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "meta/llama-guard-4-12b": {
    "model_developer": "Meta",
    "model_type": "specialized",
    "url": "https://www.llama.com/docs/model-cards-and-prompt-formats/llama-guard-4/",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.2,
      "cost_per_1M_output_tokens": 0.0,
      "source": "Together AI pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "mistral/ministral-3b-2512": {
    "model_developer": "Mistral AI",
    "model_type": "generalist",
    "url": "https://docs.mistral.ai/models/ministral-3-3b-25-12",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.1,
      "cost_per_1M_output_tokens": 0.1,
      "source": "Ministral API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "mistral/ministral-14b-2512": {
    "model_developer": "Mistral AI",
    "model_type": "generalist",
    "url": "https://docs.mistral.ai/models/ministral-3-14b-25-12",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.2,
      "cost_per_1M_output_tokens": 0.2,
      "source": "Ministral API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "mistral/mistral-large-3": {
    "model_developer": "Mistral AI",
    "model_type": "generalist",
    "url": "https://docs.mistral.ai/models/mistral-large-3-25-12",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.5,
      "cost_per_1M_output_tokens": 1.5,
      "source": "Mistral API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-5-nano": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://platform.openai.com/docs/models/gpt-5-nano",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.05,
      "cost_per_1M_output_tokens": 0.4,
      "source": "OpenAI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "Reasoning effort set to minimal; see OpenAI API docs. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-5-mini": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://platform.openai.com/docs/models/gpt-5-mini",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.25,
      "cost_per_1M_output_tokens": 2.0,
      "source": "OpenAI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "Reasoning effort set to low; see OpenAI API docs. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-5.2": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://platform.openai.com/docs/models/gpt-5.2",
    "cost_info": {
      "cost_per_1M_input_tokens": 1.75,
      "cost_per_1M_output_tokens": 14.0,
      "source": "OpenAI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "Reasoning effort set to low; see OpenAI API docs. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-5.2-informed": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://platform.openai.com/docs/models/gpt-5.2",
    "cost_info": {
      "cost_per_1M_input_tokens": 1.75,
      "cost_per_1M_output_tokens": 14.0,
      "source": "OpenAI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "Reasoning effort set to low; see OpenAI API docs. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-5.2-uninformed": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://platform.openai.com/docs/models/gpt-5.2",
    "cost_info": {
      "cost_per_1M_input_tokens": 1.75,
      "cost_per_1M_output_tokens": 14.0,
      "source": "OpenAI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "Reasoning effort set to low; see OpenAI API docs. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-5.4": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://platform.openai.com/docs/models/gpt-5.4",
    "cost_info": {
      "cost_per_1M_input_tokens": 2.5,
      "cost_per_1M_output_tokens": 22.5,
      "source": "OpenAI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "Reasoning effort set to low; see OpenAI API docs. REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-oss-20b": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://www.together.ai/models/gpt-oss-20b",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.05,
      "cost_per_1M_output_tokens": 0.2,
      "source": "Together AI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-oss-120b": {
    "model_developer": "OpenAI",
    "model_type": "generalist",
    "url": "https://www.together.ai/models/gpt-oss-120b",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.05,
      "cost_per_1M_output_tokens": 0.2,
      "source": "Together AI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/gpt-oss-safeguard-20b": {
    "model_developer": "OpenAI",
    "model_type": "specialized",
    "url": "https://huggingface.co/openai/gpt-oss-safeguard-20b",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.07,
      "cost_per_1M_output_tokens": 0.3,
      "source": "OpenRouter API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "openai/omni-moderation": {
    "model_developer": "OpenAI",
    "model_type": "specialized",
    "url": "https://platform.openai.com/docs/models/omni-moderation-latest",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "OpenAI Moderation API pricing",
      "cost_per_h": "N/A",
      "additional_info": "The Omni Moderation endpoint is free to use but has low rate limits in the free tier. Your usage tier on the OpenAI API determines these rate limits."
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "virtue-ai/virtueguard-text-lite": {
    "model_developer": "Virtue AI",
    "model_type": "specialized",
    "url": "https://www.together.ai/models/virtueguard-text-lite",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.2,
      "cost_per_1M_output_tokens": 0.0,
      "source": "Together AI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "x-ai/grok-4-1-fast-non-reasoning": {
    "model_developer": "X-AI",
    "model_type": "generalist",
    "url": "https://docs.x.ai/docs/models/grok-4-1-fast-non-reasoning",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.2,
      "cost_per_1M_output_tokens": 0.5,
      "source": "X-AI API pricing",
      "cost_per_h": "N/A",
      "additional_info": ""
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "aws/bedrock-guardrail": {
    "model_developer": "Amazon Web Services",
    "model_type": "specialized",
    "url": "https://aws.amazon.com/bedrock/guardrails/",
    "cost_info": {
      "cost_per_1M_input_tokens": 150.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "Amazon Bedrock API pricing",
      "cost_per_h": "N/A",
      "additional_info": "Usage is measured in text units. One text unit may contain up to 1000 characters. Only the input string is considered in the text unit.\n1000 text units cost $0.15. Assuming an average length of 500 characters for a message, and a token length of approx. 4 characters: Estimated cost per 1M input tokens = 1M/(500/4) text units * $0.15/1000 text units) = $1.2.\nCost per 1M input tokens (not units) = 1.2$."
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "nvidia/aegis-ai-content-safety-llamaguard-defensive-1.0": {
    "model_developer": "NVIDIA",
    "model_type": "specialized",
    "url": "https://huggingface.co/nvidia/Aegis-AI-Content-Safety-LlamaGuard-Defensive-1.0",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "google/shieldgemma-2b": {
    "model_developer": "Google",
    "model_type": "specialized",
    "url": "https://huggingface.co/google/shieldgemma-2b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "google/shieldgemma-9b": {
    "model_developer": "Google",
    "url": "https://huggingface.co/google/shieldgemma-9b",
    "model_type": "specialized",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "google/shieldgemma-27b": {
    "model_developer": "Google",
    "model_type": "specialized",
    "url": "https://huggingface.co/google/shieldgemma-27b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "saillab/xguard": {
    "model_developer": "SAIL Lab",
    "model_type": "specialized",
    "url": "https://huggingface.co/saillab/x-guard",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "openai/gpt-oss-safeguard-120b": {
    "model_developer": "OpenAI",
    "model_type": "specialized",
    "url": "https://huggingface.co/openai/gpt-oss-safeguard-120b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 3.07,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (94GB NVL) on RunPod using vLLM."
    }
  },
  "qwen3guard-gen-8b": {
    "model_developer": "Qwen",
    "model_type": "specialized",
    "url": "https://huggingface.co/Qwen/Qwen3Guard-Gen-8B",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "qwen/qwen3guard-gen-4b": {
    "model_developer": "Qwen",
    "model_type": "specialized",
    "url": "https://huggingface.co/Qwen/Qwen3Guard-Gen-4B",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "qwen/qwen3guard-gen-0.6b": {
    "model_developer": "Qwen",
    "model_type": "specialized",
    "url": "https://huggingface.co/Qwen/Qwen3Guard-Gen-0.6B",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "nvidia/llama-3.1-nemotron-safety-guard-8b-v3": {
    "model_developer": "NVIDIA",
    "model_type": "specialized",
    "url": "https://huggingface.co/nvidia/Llama-3.1-Nemotron-Safety-Guard-8B-v3",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "rakancorle1/thinkguard": {
    "model_developer": "RakanCorle1",
    "model_type": "specialized",
    "url": "https://huggingface.co/Rakancorle1/ThinkGuard",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "allenai/wildguard": {
    "model_developer": "AllenAI",
    "model_type": "specialized",
    "url": "https://huggingface.co/allenai/wildguard",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "toxicityprompts/polyguard-ministral": {
    "model_developer": "ToxicityPrompts",
    "model_type": "specialized",
    "url": "https://huggingface.co/ToxicityPrompts/PolyGuard-Ministral",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "toxicityprompts/polyguard-qwen": {
    "model_developer": "ToxicityPrompts",
    "model_type": "specialized",
    "url": "https://huggingface.co/ToxicityPrompts/PolyGuard-Qwen",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "toxicityprompts/polyguard-qwen-smol": {
    "model_developer": "ToxicityPrompts",
    "model_type": "specialized",
    "url": "https://huggingface.co/ToxicityPrompts/PolyGuard-Qwen-Smol",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.0-2b": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.0-2b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.0-8b": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.0-8b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.1-2b": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.1-2b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.1-8b": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.1-8b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.2-5b": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.2-5b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.2-3b-a800m": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.2-3b-a800m",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "ibm-granite/granite-guardian-3.3-8b": {
    "model_developer": "IBM",
    "model_type": "specialized",
    "url": "https://huggingface.co/ibm-granite/granite-guardian-3.3-8b",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "govtech/lionguard-2": {
    "model_developer": "GovTech",
    "model_type": "specialized",
    "url": "https://huggingface.co/govtech/lionguard-2",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the transformers Library."
    }
  },
  "govtech/lionguard-2.1": {
    "model_developer": "GovTech",
    "model_type": "specialized",
    "url": "https://huggingface.co/govtech/lionguard-2.1",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the transformers Library."
    }
  },
  "govtech/lionguard-2-lite": {
    "model_developer": "GovTech",
    "model_type": "specialized",
    "url": "https://huggingface.co/govtech/lionguard-2-lite",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the transformers Library."
    }
  },
  "leolee99/piguard": {
    "model_developer": "Leolee99",
    "model_type": "specialized",
    "url": "https://huggingface.co/leolee99/PIGuard",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the transformers Library."
    }
  },
  "meta-llama/llama-prompt-guard-2-22m": {
    "model_developer": "Meta",
    "model_type": "specialized",
    "url": "https://huggingface.co/meta-llama/Llama-Prompt-Guard-2-22M",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the transformers Library."
    }
  },
  "meta-llama/llama-prompt-guard-2-86m": {
    "model_developer": "Meta",
    "model_type": "specialized",
    "url": "https://huggingface.co/meta-llama/Llama-Prompt-Guard-2-86M",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the transformers Library."
    }
  },
  "protectai/llm-guard": {
    "model_developer": "ProtectAI",
    "model_type": "specialized",
    "url": "https://protectai.github.io/llm-guard/input_scanners/prompt_injection/",
    "cost_info": {
      "cost_per_1M_input_tokens": "N/A",
      "cost_per_1M_output_tokens": "N/A",
      "source": "RunPod",
      "cost_per_h": 2.39,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an H100 (80GB PCIe) on RunPod using the llm-guard Python SDK."
    }
  },
  "neuraltrust/promptguard": {
    "model_developer": "NeuralTrust",
    "model_type": "specialized",
    "url": "https://neuraltrust.ai/prompt-guard",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "NeuralTrust",
      "cost_per_h": "N/A",
      "additional_info": "Usage is free on the API. We were banned after about 1800 requests."
    },
    "execution_specifications": {
      "type": "API",
      "details": "REST API accessed through a CPU3 RunPod instance at US-KS-2."
    }
  },
  "bells-o/opencc-jb-escalation": {
    "model_developer": "CeSIA",
    "model_type": "specialized",
    "url": "https://huggingface.co/centrepourlasecuriteia/opencc-jb-escalation",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "RunPod",
      "cost_per_h": 1.49,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an A100 (80GB PCIe) on RunPod using vLLM."
    }
  },
  "bells-o/opencc-cm-escalation": {
    "model_developer": "CeSIA",
    "model_type": "specialized",
    "url": "https://huggingface.co/centrepourlasecuriteia/opencc-cm-escalation",
    "cost_info": {
      "cost_per_1M_input_tokens": 0.0,
      "cost_per_1M_output_tokens": 0.0,
      "source": "RunPod",
      "cost_per_h": 1.49,
      "additional_info": "Output token cost is disregarded. The cost per 1M input tokens is estimated as total_cost * (1,000,000 / total_input_tokens)."
    },
    "execution_specifications": {
      "type": "Local",
      "details": "This model was ran on an A100 (80GB PCIe) on RunPod using vLLM."
    }
  }
}