{
  "timestamp": "20260718_214307",
  "generated_by": "tokenizer_bench",
  "dataset_groups": {
    "C++": "Code",
    "Javascript": "Code",
    "Python": "Code",
    "Rust": "Code",
    "Go": "Code",
    "Mandarin": "Multilingual",
    "English": "Multilingual",
    "German": "Multilingual",
    "French": "Multilingual",
    "Arabic": "Multilingual",
    "Spanish": "Multilingual",
    "Hermes Agent Traces": "Agentic Traces",
    "perfectblend": "Perfect Blend"
  },
  "models": {
    "Qwen3.6-27B": {
      "model_id": "Qwen3.6-27B",
      "model_name": "Qwen3.6-27B",
      "release_date": "April 2026",
      "vocab_size": 248044,
      "per_million_output_token_price": 2.7,
      "throughput": 30,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9149932,
          "chars_per_token": 3.3068
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 36291899,
          "chars_per_token": 3.3378
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6691160,
          "chars_per_token": 4.368
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7093615,
          "chars_per_token": 4.0595
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 6872540,
          "chars_per_token": 4.1421
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27804267,
          "chars_per_token": 3.4556
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 37195280,
          "chars_per_token": 2.9117
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 36953764,
          "chars_per_token": 2.8972
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 9563690,
          "chars_per_token": 1.6141
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 18610232,
          "chars_per_token": 3.5874
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 30517992,
          "chars_per_token": 3.5084
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7196442,
          "chars_per_token": 4.3309
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 481128,
          "chars_per_token": 4.1226
        }
      }
    },
    "DeepSeek-V4-Pro": {
      "model_id": "DeepSeek-V4-Pro",
      "model_name": "DeepSeek-V4-Pro",
      "release_date": "April 2026",
      "vocab_size": 128000,
      "per_million_output_token_price": 0.87,
      "throughput": 46,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 11295235,
          "chars_per_token": 2.6787
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 37611840,
          "chars_per_token": 3.2207
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6472399,
          "chars_per_token": 4.5156
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7782233,
          "chars_per_token": 3.7003
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 7777627,
          "chars_per_token": 3.6601
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 28978858,
          "chars_per_token": 3.3155
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 34089412,
          "chars_per_token": 3.177
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 37129312,
          "chars_per_token": 2.8835
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 9740840,
          "chars_per_token": 1.5847
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 19410672,
          "chars_per_token": 3.4395
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 32404495,
          "chars_per_token": 3.3042
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 8029497,
          "chars_per_token": 3.8816
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 465756,
          "chars_per_token": 4.2587
        }
      }
    },
    "gemma-4-31B-it": {
      "model_id": "gemma-4-31B-it",
      "model_name": "Gemma-4-31B-it",
      "release_date": "March 2026",
      "vocab_size": 262144,
      "per_million_output_token_price": 0.97,
      "throughput": 46,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 10203325,
          "chars_per_token": 2.9654
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 39984775,
          "chars_per_token": 3.0295
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6509849,
          "chars_per_token": 4.4896
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7111295,
          "chars_per_token": 4.0494
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 6935367,
          "chars_per_token": 4.1045
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 30370870,
          "chars_per_token": 3.1635
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 40710370,
          "chars_per_token": 2.6603
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 41305005,
          "chars_per_token": 2.592
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 11013420,
          "chars_per_token": 1.4016
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 21303112,
          "chars_per_token": 3.134
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 33843088,
          "chars_per_token": 3.1637
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7045705,
          "chars_per_token": 4.4235
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 482328,
          "chars_per_token": 4.1124
        }
      }
    },
    "Inkling": {
      "model_id": "Inkling",
      "model_name": "Inkling",
      "release_date": "July 2026",
      "vocab_size": 199998,
      "per_million_output_token_price": 4.05,
      "throughput": 50,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9446095,
          "chars_per_token": 3.2031
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 34900692,
          "chars_per_token": 3.4709
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6444531,
          "chars_per_token": 4.5351
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 6812404,
          "chars_per_token": 4.2271
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 6675677,
          "chars_per_token": 4.2642
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27123903,
          "chars_per_token": 3.5422
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 32091996,
          "chars_per_token": 3.3747
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 34719133,
          "chars_per_token": 3.0837
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 12069020,
          "chars_per_token": 1.279
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 17985169,
          "chars_per_token": 3.7121
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 29890158,
          "chars_per_token": 3.5821
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7013962,
          "chars_per_token": 4.4436
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 462547,
          "chars_per_token": 4.2882
        }
      }
    },
    "Hy3": {
      "model_id": "Hy3",
      "model_name": "Hy3",
      "release_date": "July 2026",
      "vocab_size": 120000,
      "per_million_output_token_price": 0.8,
      "throughput": 24,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 13908409,
          "chars_per_token": 2.1754
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 36291403,
          "chars_per_token": 3.3379
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6557259,
          "chars_per_token": 4.4572
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 8418521,
          "chars_per_token": 3.4206
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 8646184,
          "chars_per_token": 3.2924
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 28124549,
          "chars_per_token": 3.4162
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 33485062,
          "chars_per_token": 3.2343
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 36452441,
          "chars_per_token": 2.9371
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 10001926,
          "chars_per_token": 1.5434
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 18691341,
          "chars_per_token": 3.5719
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 31321304,
          "chars_per_token": 3.4184
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 9063194,
          "chars_per_token": 3.4389
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 465820,
          "chars_per_token": 4.2581
        }
      }
    },
    "GLM-5.2": {
      "model_id": "GLM-5.2",
      "model_name": "GLM-5.2",
      "release_date": "June 2026",
      "vocab_size": 154820,
      "per_million_output_token_price": 4.4,
      "throughput": 32,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 12218259,
          "chars_per_token": 2.4763
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 35023095,
          "chars_per_token": 3.4587
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6514700,
          "chars_per_token": 4.4863
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7542899,
          "chars_per_token": 3.8177
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 7379737,
          "chars_per_token": 3.8574
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27048858,
          "chars_per_token": 3.5521
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 33344182,
          "chars_per_token": 3.248
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 35165042,
          "chars_per_token": 3.0446
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 10241910,
          "chars_per_token": 1.5072
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 18022971,
          "chars_per_token": 3.7043
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 29869683,
          "chars_per_token": 3.5846
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7726763,
          "chars_per_token": 4.0336
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 467568,
          "chars_per_token": 4.2422
        }
      }
    },
    "LongCat-2.0": {
      "model_id": "LongCat-2.0",
      "model_name": "LongCat-2.0",
      "release_date": "July 2026",
      "vocab_size": 163840,
      "per_million_output_token_price": 2.95,
      "throughput": null,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 15607669,
          "chars_per_token": 1.9386
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 36765855,
          "chars_per_token": 3.2948
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6694720,
          "chars_per_token": 4.3656
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7964282,
          "chars_per_token": 3.6157
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 8453518,
          "chars_per_token": 3.3674
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27753827,
          "chars_per_token": 3.4618
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 37057555,
          "chars_per_token": 2.9225
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 37236815,
          "chars_per_token": 2.8752
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 9465791,
          "chars_per_token": 1.6308
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 18882621,
          "chars_per_token": 3.5357
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 30749893,
          "chars_per_token": 3.482
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 8312500,
          "chars_per_token": 3.7494
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 483456,
          "chars_per_token": 4.1028
        }
      }
    },
    "NVIDIA-Nemotron-3-Ultra-550B": {
      "model_id": "NVIDIA-Nemotron-3-Ultra-550B",
      "model_name": "NVIDIA-Nemotron-3-Ultra-550B",
      "release_date": "June 2026",
      "vocab_size": 131072,
      "per_million_output_token_price": 2.2,
      "throughput": 132,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9122321,
          "chars_per_token": 3.3168
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 38634762,
          "chars_per_token": 3.1354
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6731647,
          "chars_per_token": 4.3417
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7024546,
          "chars_per_token": 4.0994
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 7176157,
          "chars_per_token": 3.9668
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 29369704,
          "chars_per_token": 3.2714
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 38689207,
          "chars_per_token": 2.7993
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 39429614,
          "chars_per_token": 2.7153
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 15036125,
          "chars_per_token": 1.0266
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 19807829,
          "chars_per_token": 3.3705
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 32333528,
          "chars_per_token": 3.3114
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7450397,
          "chars_per_token": 4.1833
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 488158,
          "chars_per_token": 4.0633
        }
      }
    },
    "Kimi-K3": {
      "model_id": "Kimi-K3",
      "model_name": "Kimi-K3",
      "release_date": "June 2026",
      "vocab_size": 163840,
      "per_million_output_token_price": 15.0,
      "throughput": 33,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 14346510,
          "chars_per_token": 2.109
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 34706925,
          "chars_per_token": 3.4902
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6532956,
          "chars_per_token": 4.4737
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 8646128,
          "chars_per_token": 3.3306
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 8800416,
          "chars_per_token": 3.2347
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27013111,
          "chars_per_token": 3.5568
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 31618277,
          "chars_per_token": 3.4253
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 34762845,
          "chars_per_token": 3.0798
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 9558444,
          "chars_per_token": 1.615
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 17997308,
          "chars_per_token": 3.7096
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 29746175,
          "chars_per_token": 3.5994
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 9123558,
          "chars_per_token": 3.4161
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 464057,
          "chars_per_token": 4.2743
        }
      }
    },
    "MiniMax-M3": {
      "model_id": "MiniMax-M3",
      "model_name": "MiniMax-M3",
      "release_date": "June 2026",
      "vocab_size": 200000,
      "per_million_output_token_price": 1.2,
      "throughput": 51,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9198594,
          "chars_per_token": 3.2893
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 35614925,
          "chars_per_token": 3.4013
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6372086,
          "chars_per_token": 4.5867
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 6508428,
          "chars_per_token": 4.4245
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 7563502,
          "chars_per_token": 3.7637
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27374674,
          "chars_per_token": 3.5098
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 31849922,
          "chars_per_token": 3.4004
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 35612678,
          "chars_per_token": 3.0063
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 9322596,
          "chars_per_token": 1.6558
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 18273721,
          "chars_per_token": 3.6535
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 30159419,
          "chars_per_token": 3.5501
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 6671461,
          "chars_per_token": 4.6717
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 459923,
          "chars_per_token": 4.3127
        }
      }
    },
    "gpt-5.6 Sol": {
      "model_id": "gpt-5.6 Sol",
      "model_name": "GPT-5.6 Sol",
      "release_date": "July 2026",
      "vocab_size": 200019,
      "per_million_output_token_price": 30,
      "throughput": 42,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9446095,
          "chars_per_token": 3.2031
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 34900692,
          "chars_per_token": 3.4709
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6444531,
          "chars_per_token": 4.5351
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 6812404,
          "chars_per_token": 4.2271
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 6675677,
          "chars_per_token": 4.2642
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 27123903,
          "chars_per_token": 3.5422
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 32091996,
          "chars_per_token": 3.3747
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 34719133,
          "chars_per_token": 3.0837
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 12069020,
          "chars_per_token": 1.279
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 17985169,
          "chars_per_token": 3.7121
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 29890158,
          "chars_per_token": 3.5821
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7013962,
          "chars_per_token": 4.4436
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 462547,
          "chars_per_token": 4.2882
        }
      }
    },
    "gemini-3.5-flash": {
      "model_id": "gemini-3.5-flash",
      "model_name": "Gemini-3.5-Flash",
      "release_date": "May 2026",
      "vocab_size": null,
      "per_million_output_token_price": 9.0,
      "throughput": 125,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 10213309,
          "chars_per_token": 2.9625
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 39992636,
          "chars_per_token": 3.0289
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6574572,
          "chars_per_token": 4.4454
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7121270,
          "chars_per_token": 4.0437
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 6945358,
          "chars_per_token": 4.0986
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 30379080,
          "chars_per_token": 3.1627
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 40703206,
          "chars_per_token": 2.6608
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 41266651,
          "chars_per_token": 2.5944
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 11023414,
          "chars_per_token": 1.4003
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 21302103,
          "chars_per_token": 3.1341
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 33803932,
          "chars_per_token": 3.1674
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7055696,
          "chars_per_token": 4.4173
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 483144,
          "chars_per_token": 4.1054
        }
      }
    },
    "grok-4.5": {
      "model_id": "grok-4.5",
      "model_name": "Grok-4.5",
      "release_date": "June 2026",
      "vocab_size": null,
      "per_million_output_token_price": 6.0,
      "throughput": 51,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9251713,
          "chars_per_token": 3.2704
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 37746960,
          "chars_per_token": 3.2091
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6443230,
          "chars_per_token": 4.536
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7151473,
          "chars_per_token": 4.0267
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 6797524,
          "chars_per_token": 4.1878
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 29403423,
          "chars_per_token": 3.2676
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 34653389,
          "chars_per_token": 3.1253
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 38546147,
          "chars_per_token": 2.7776
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 10751790,
          "chars_per_token": 1.4357
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 20174863,
          "chars_per_token": 3.3092
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 32470387,
          "chars_per_token": 3.2975
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7169677,
          "chars_per_token": 4.3471
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 465864,
          "chars_per_token": 4.2577
        }
      }
    },
    "claude-fable-5": {
      "model_id": "claude-fable-5",
      "model_name": "Claude-Fable-5",
      "release_date": "June 2026",
      "vocab_size": null,
      "per_million_output_token_price": 50.0,
      "throughput": 48,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 19799576,
          "chars_per_token": 1.5281
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 58819248,
          "chars_per_token": 2.0595
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 10166801,
          "chars_per_token": 2.8747
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 11821597,
          "chars_per_token": 2.4359
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 13816630,
          "chars_per_token": 2.0603
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 45174889,
          "chars_per_token": 2.1268
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 51190029,
          "chars_per_token": 2.1157
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 54354727,
          "chars_per_token": 1.9697
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 17179847,
          "chars_per_token": 0.8985
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 29120289,
          "chars_per_token": 2.2927
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 49172923,
          "chars_per_token": 2.1774
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 12479626,
          "chars_per_token": 2.4974
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 689048,
          "chars_per_token": 2.8786
        }
      }
    },
    "LiquidAI/LFM2.5-8B-A1B": {
      "model_id": "LiquidAI/LFM2.5-8B-A1B",
      "model_name": "LFM2.5-8B-A1B",
      "release_date": "May 2026",
      "vocab_size": 124893,
      "per_million_output_token_price": null,
      "throughput": null,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 9306910,
          "chars_per_token": 3.251
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 37706605,
          "chars_per_token": 3.2126
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 6612549,
          "chars_per_token": 4.4199
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 7243492,
          "chars_per_token": 3.9755
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 7161701,
          "chars_per_token": 3.9748
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 29220284,
          "chars_per_token": 3.2881
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 33917174,
          "chars_per_token": 3.1931
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 36572359,
          "chars_per_token": 2.9275
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 12327229,
          "chars_per_token": 1.2522
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 19446956,
          "chars_per_token": 3.4331
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 32531813,
          "chars_per_token": 3.2912
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 7616880,
          "chars_per_token": 4.0918
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 472281,
          "chars_per_token": 4.1999
        }
      }
    },
    "claude-opus-4.6": {
      "model_id": "claude-opus-4.6",
      "model_name": "Claude-Opus-4.6",
      "release_date": "February 2026",
      "vocab_size": null,
      "per_million_output_token_price": 25,
      "throughput": 34,
      "datasets": {
        "Arabic": {
          "dataset_name": "Arabic",
          "total_chars": 30256625,
          "total_tokens": 19692764,
          "chars_per_token": 1.5364
        },
        "C++": {
          "dataset_name": "C++",
          "total_chars": 121135642,
          "total_tokens": 43184920,
          "chars_per_token": 2.805
        },
        "English": {
          "dataset_name": "English",
          "total_chars": 29226784,
          "total_tokens": 7224971,
          "chars_per_token": 4.0452
        },
        "French": {
          "dataset_name": "French",
          "total_chars": 28796491,
          "total_tokens": 8958106,
          "chars_per_token": 3.2146
        },
        "German": {
          "dataset_name": "German",
          "total_chars": 28466537,
          "total_tokens": 9846506,
          "chars_per_token": 2.891
        },
        "Go": {
          "dataset_name": "Go",
          "total_chars": 96079303,
          "total_tokens": 33515550,
          "chars_per_token": 2.8667
        },
        "Hermes Agent Traces": {
          "dataset_name": "Hermes Agent Traces",
          "total_chars": 108301661,
          "total_tokens": 38665060,
          "chars_per_token": 2.801
        },
        "Javascript": {
          "dataset_name": "Javascript",
          "total_chars": 107064159,
          "total_tokens": 43122279,
          "chars_per_token": 2.4828
        },
        "Mandarin": {
          "dataset_name": "Mandarin",
          "total_chars": 15436567,
          "total_tokens": 17029884,
          "chars_per_token": 0.9064
        },
        "Python": {
          "dataset_name": "Python",
          "total_chars": 66763191,
          "total_tokens": 22187993,
          "chars_per_token": 3.009
        },
        "Rust": {
          "dataset_name": "Rust",
          "total_chars": 107069705,
          "total_tokens": 36791819,
          "chars_per_token": 2.9101
        },
        "Spanish": {
          "dataset_name": "Spanish",
          "total_chars": 31166987,
          "total_tokens": 9213900,
          "chars_per_token": 3.3826
        },
        "perfectblend": {
          "dataset_name": "perfectblend",
          "total_chars": 1983515,
          "total_tokens": 512379,
          "chars_per_token": 3.8712
        }
      }
    }
  }
}