{
 "methodology_version": "2.0",
 "generated_at": "2026-10-30T00:44:27Z",
 "as_of": "2026-10-29T00:00:00Z",
 "valid_until": "2026-10-31T00:00:00Z",
 "window": {
  "start": "2026-10-22T00:00:00Z",
  "end": "2026-10-29T00:00:00Z",
  "days": 7,
  "observed_days": 7.0
 },
 "model": {
  "id": "nvidia/nemotron-3-ultra-550b-a55b",
  "permaslug": "nvidia/nemotron-3-ultra-550b-a55b-20260604",
  "name": "Nemotron 3 Ultra",
  "maker": "NVIDIA",
  "file": "models/nemotron-3-ultra-550b-a55b.json"
 },
 "metric": "invalid_tool_call_rate",
 "status": "ok",
 "status_reason": null,
 "confidence": "high",
 "recommendation": {
  "order": [
   "deepinfra/fp4/flex",
   "venice/fp8",
   "deepinfra/fp4"
  ],
  "ignore": [
   "baseten/fp4"
  ]
 },
 "thresholds": {
  "min_tool_calls_per_day_to_recommend": 1000,
  "median_invalid_tool_call_rate": 0.010733,
  "ignore_at_or_above": 0.03
 },
 "choices": [
  {
   "id": "most_reliable",
   "also": [
    "lowest_cost"
   ],
   "label": "Most reliable, lowest cost",
   "criterion": "the lowest invalid tool-call rate; the lowest price, blended at 3 tokens in to 1 out",
   "provider": "deepinfra/fp4/flex",
   "recommendation": {
    "order": [
     "deepinfra/fp4/flex",
     "venice/fp8",
     "deepinfra/fp4"
    ],
    "ignore": [
     "baseten/fp4"
    ]
   }
  },
  {
   "id": "highest_throughput",
   "also": [],
   "label": "Highest throughput",
   "criterion": "the highest median tokens per second",
   "provider": "venice/fp8",
   "recommendation": {
    "order": [
     "venice/fp8",
     "deepinfra/fp4/flex",
     "deepinfra/fp4"
    ],
    "ignore": [
     "baseten/fp4"
    ]
   }
  },
  {
   "id": "highest_uptime",
   "also": [],
   "label": "Highest uptime",
   "criterion": "the highest uptime",
   "provider": "deepinfra/fp4",
   "recommendation": {
    "order": [
     "deepinfra/fp4",
     "deepinfra/fp4/flex",
     "venice/fp8"
    ],
    "ignore": [
     "baseten/fp4"
    ]
   }
  }
 ],
 "providers": [
  {
   "id": "deepinfra/fp4/flex",
   "name": "DeepInfra",
   "rank": 1,
   "role": "recommended",
   "slugs": [
    "deepinfra/fp4/flex"
   ],
   "variant": "fp4/flex",
   "makers_api": false,
   "retains_prompts": false,
   "trains_on_prompts": false,
   "sample_size": 11192,
   "failures": 19,
   "invalid_tool_call_rate": 0.001698,
   "invalid_tool_call_rate_ci95": [
    0.001087,
    0.00265
   ],
   "tool_calls_per_day": 1599,
   "input_price_per_million": 0.4,
   "output_price_per_million": 1.76,
   "tokens_per_second": null,
   "uptime": 0.9953,
   "quantizations": [
    "fp4"
   ],
   "endpoints": [
    {
     "tag": "deepinfra/fp4/flex",
     "tier": false,
     "name": "DeepInfra",
     "quantization": "fp4",
     "sample_size": 11192,
     "failures": 19,
     "invalid_tool_call_rate": 0.001698
    }
   ],
   "direct_api": {
    "model_id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
    "dedicated_endpoint": false,
    "context_tokens": 262144,
    "tool_choice": true,
    "docs": "https://docs.deepinfra.com/chat/tool-calling",
    "website": "https://deepinfra.com",
    "status_page": "https://status.deepinfra.com/",
    "terms": "https://deepinfra.com/terms",
    "privacy": "https://deepinfra.com/privacy",
    "openrouter": "https://openrouter.ai/provider/deepinfra"
   }
  },
  {
   "id": "venice/fp8",
   "name": "Venice",
   "rank": 2,
   "role": "recommended",
   "slugs": [
    "venice/fp8"
   ],
   "variant": "fp8",
   "makers_api": false,
   "retains_prompts": false,
   "trains_on_prompts": false,
   "sample_size": 80352,
   "failures": 151,
   "invalid_tool_call_rate": 0.001879,
   "invalid_tool_call_rate_ci95": [
    0.001603,
    0.002203
   ],
   "tool_calls_per_day": 11479,
   "input_price_per_million": 0.625,
   "output_price_per_million": 3.125,
   "tokens_per_second": 94.0,
   "uptime": 0.8161,
   "quantizations": [
    "fp8"
   ],
   "endpoints": [
    {
     "tag": "venice/fp8",
     "tier": false,
     "name": "Venice",
     "quantization": "fp8",
     "sample_size": 80352,
     "failures": 151,
     "invalid_tool_call_rate": 0.001879
    }
   ],
   "direct_api": {
    "model_id": "nvidia-nemotron-3-ultra-550b-a55b",
    "dedicated_endpoint": false,
    "context_tokens": 256000,
    "tool_choice": true,
    "docs": "https://docs.venice.ai/overview/about-venice",
    "website": "https://venice.ai",
    "status_page": null,
    "terms": "https://venice.ai/legal/tos",
    "privacy": "https://venice.ai/legal/privacy-policy",
    "openrouter": "https://openrouter.ai/provider/venice"
   }
  },
  {
   "id": "deepinfra/fp4",
   "name": "DeepInfra",
   "rank": 3,
   "role": "recommended",
   "slugs": [
    "deepinfra/fp4"
   ],
   "variant": "fp4",
   "makers_api": false,
   "retains_prompts": false,
   "trains_on_prompts": false,
   "sample_size": 54134,
   "failures": 581,
   "invalid_tool_call_rate": 0.010733,
   "invalid_tool_call_rate_ci95": [
    0.009899,
    0.011636
   ],
   "tool_calls_per_day": 7733,
   "input_price_per_million": 0.5,
   "output_price_per_million": 2.2,
   "tokens_per_second": 88.0,
   "uptime": 0.9989,
   "quantizations": [
    "fp4"
   ],
   "endpoints": [
    {
     "tag": "deepinfra/fp4",
     "tier": false,
     "name": "DeepInfra",
     "quantization": "fp4",
     "sample_size": 54134,
     "failures": 581,
     "invalid_tool_call_rate": 0.010733
    }
   ],
   "direct_api": {
    "model_id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
    "dedicated_endpoint": false,
    "context_tokens": 262144,
    "tool_choice": true,
    "docs": "https://docs.deepinfra.com/chat/tool-calling",
    "website": "https://deepinfra.com",
    "status_page": "https://status.deepinfra.com/",
    "terms": "https://deepinfra.com/terms",
    "privacy": "https://deepinfra.com/privacy",
    "openrouter": "https://openrouter.ai/provider/deepinfra"
   }
  },
  {
   "id": "baseten/fp4",
   "name": "Baseten",
   "rank": 4,
   "role": "ignored",
   "slugs": [
    "baseten/fp4"
   ],
   "variant": "fp4",
   "makers_api": false,
   "retains_prompts": false,
   "trains_on_prompts": false,
   "sample_size": 134462,
   "failures": 4817,
   "invalid_tool_call_rate": 0.035824,
   "invalid_tool_call_rate_ci95": [
    0.034844,
    0.036831
   ],
   "tool_calls_per_day": 19209,
   "input_price_per_million": 0.6,
   "output_price_per_million": 2.4,
   "tokens_per_second": 105.0,
   "uptime": 0.9396,
   "quantizations": [
    "fp4"
   ],
   "endpoints": [
    {
     "tag": "baseten/fp4",
     "tier": false,
     "name": "Baseten",
     "quantization": "fp4",
     "sample_size": 40997,
     "failures": 737,
     "invalid_tool_call_rate": 0.017977
    },
    {
     "tag": "baseten/fp4/us",
     "tier": false,
     "name": "Baseten",
     "quantization": "fp4",
     "sample_size": 93465,
     "failures": 4080,
     "invalid_tool_call_rate": 0.043653
    }
   ],
   "direct_api": {
    "model_id": "nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B",
    "dedicated_endpoint": false,
    "context_tokens": 202800,
    "tool_choice": true,
    "docs": "https://docs.baseten.co/inference/model-apis/overview",
    "website": "https://www.baseten.co",
    "status_page": "https://status.baseten.co/",
    "terms": "https://www.baseten.co/terms-and-conditions",
    "privacy": "https://www.baseten.co/privacy-policy",
    "openrouter": "https://openrouter.ai/provider/baseten"
   }
  }
 ],
 "others": [],
 "structured_outputs": {
  "metric": "structured_output_error_rate",
  "as_of": "2026-10-29T00:00:00Z",
  "valid_until": "2026-10-31T00:00:00Z",
  "window": {
   "start": "2026-10-22T00:00:00Z",
   "end": "2026-10-29T00:00:00Z",
   "days": 7,
   "observed_days": 7.0
  },
  "status": "single_provider",
  "status_reason": "fewer than two providers at 1,000 structured outputs a day or more",
  "confidence": "none",
  "recommendation": null,
  "thresholds": {
   "min_structured_outputs_per_day_to_recommend": 1000,
   "median_structured_output_error_rate": null,
   "ignore_at_or_above": 0.03
  },
  "providers": [],
  "others": [
   {
    "id": "deepinfra/fp4/flex",
    "name": "DeepInfra",
    "role": "neutral",
    "slugs": [
     "deepinfra/fp4/flex"
    ],
    "variant": "fp4/flex",
    "sample_size": 360,
    "failures": 1,
    "structured_output_error_rate": 0.002778,
    "structured_output_error_rate_ci95": [
     0.000491,
     0.015565
    ],
    "structured_outputs_per_day": 51,
    "endpoints": [
     {
      "tag": "deepinfra/fp4/flex",
      "tier": false,
      "sample_size": 360,
      "failures": 1,
      "structured_output_error_rate": 0.002778
     }
    ]
   },
   {
    "id": "deepinfra/fp4",
    "name": "DeepInfra",
    "role": "neutral",
    "slugs": [
     "deepinfra/fp4"
    ],
    "variant": "fp4",
    "sample_size": 175859,
    "failures": 1363,
    "structured_output_error_rate": 0.007751,
    "structured_output_error_rate_ci95": [
     0.007351,
     0.008171
    ],
    "structured_outputs_per_day": 25123,
    "endpoints": [
     {
      "tag": "deepinfra/fp4",
      "tier": false,
      "sample_size": 175859,
      "failures": 1363,
      "structured_output_error_rate": 0.007751
     }
    ]
   }
  ]
 },
 "source": "dottxt Tool-call Index (https://dottxt.ai), from OpenRouter's published measurements"
}
