File size: 5,070 Bytes
3eed0d7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
{
  "dataset": "benchmarks-llm",
  "version": "1.0.0",
  "updated": "2026-08-12",
  "sources": [
    "https://api-docs.deepseek.com/updates/",
    "https://api-docs.deepseek.com/quick_start/pricing/",
    "https://artificialanalysis.ai/models/comparisons/deepseek-v4-flash-vs-mistral-medium-3-5",
    "https://artificialanalysis.ai/models/mistral-large-3",
    "https://benchlm.ai/best/mistral-models",
    "https://mistral.ai/pricing/",
    "https://developer.nvidia.com/blog/nvidia-nemotron-3-5-lightning-delivers-fast-accurate-specialized-task-execution-for-long-running-agents/",
    "https://openrouter.ai/nvidia/nemotron-3.5-lightning:free",
    "Rapport interne titoune33 (11-12/08/2026)"
  ],
  "license": "CC BY 4.0 (données factuelles ; rapports internes titoune33)",
  "models": [
    {
      "id": "deepseek-v4-flash-0731",
      "vendor": "DeepSeek",
      "release_date": "2026-07-31",
      "architecture": "MoE",
      "params_total_b": 284,
      "params_active_b": 13,
      "context_tokens": 1000000,
      "input_image": false,
      "open_weights": true,
      "license": "MIT",
      "price_input_usd_per_mtok": 0.14,
      "price_output_usd_per_mtok": 0.28,
      "price_cache_hit_usd_per_mtok": 0.0028,
      "intelligence_aa_index": 52,
      "speed_output_tok_per_s": 131,
      "time_to_first_token_s": 1.44,
      "agent_benchmarks": {
        "terminal_bench_2_1": 82.7,
        "nl2repo": 54.2,
        "cybergym": 76.7,
        "deepswe": 54.4,
        "toolathlon": 70.3,
        "agent_last_exam": 25.2,
        "automation_bench": 25.1
      },
      "notes": "Release officielle remplaçant la preview d'avril 2026. DeepSeek-V4-Pro annoncé."
    },
    {
      "id": "mistral-large-3",
      "vendor": "Mistral",
      "release_date": "2025-12-01",
      "architecture": "MoE",
      "params_total_b": 675,
      "params_active_b": 41,
      "context_tokens": 262000,
      "input_image": true,
      "open_weights": false,
      "license": "proprietary",
      "price_input_usd_per_mtok": 0.5,
      "price_output_usd_per_mtok": 1.5,
      "benchalign_v5": 49.4,
      "benchalign_multimodal": 67,
      "benchalign_math": 65,
      "intelligence_aa_index": 16,
      "notes": "Index AA mesuré déc. 2025 sur une version antérieure de l'index. Fort en prose française (réputation)."
    },
    {
      "id": "mistral-medium-3-5",
      "vendor": "Mistral",
      "release_date": "2026-04-01",
      "architecture": "MoE",
      "params_total_b": 128,
      "context_tokens": 256000,
      "input_image": true,
      "open_weights": true,
      "license": "modified-open",
      "price_input_usd_per_mtok": 1.16,
      "price_output_usd_per_mtok": 1.16,
      "intelligence_aa_index": 30,
      "speed_output_tok_per_s": 147,
      "notes": "Modèle par défaut de Vibe (ex-Le Chat). ~20x plus cher que DeepSeek V4 Flash."
    },
    {
      "id": "mistral-small-4",
      "vendor": "Mistral",
      "release_date": "2026-01-01",
      "architecture": null,
      "context_tokens": 256000,
      "intelligence_aa_index": 20,
      "notes": "Modèle d'entrée de gamme."
    },
    {
      "id": "nvidia-nemotron-3-5-lightning",
      "vendor": "NVIDIA",
      "release_date": "2026-08-11",
      "architecture": "MoE",
      "params_total_b": 30,
      "params_active_b": 3,
      "context_tokens": 1000000,
      "max_output_tokens": 64000,
      "intelligence_aa_index": 24,
      "price_input_usd_per_mtok": 0.05,
      "price_output_usd_per_mtok": 0.20,
      "pinchbench_accuracy_pct": 86,
      "notes": "Modèle d'exécution pour agents longue durée. Local : DGX Spark, RTX 5090, Jetson, llama.cpp. Gratuit sur OpenRouter (:free, ~20 RPM / 200 req/jour). Recommandé comme sous-agent d'un modèle frontalier."
    },
    {
      "id": "mimo-v2-5",
      "vendor": "Xiaomi",
      "release_date": "2026-04-22",
      "architecture": "MoE",
      "params_total_b": 310,
      "params_active_b": 15,
      "context_tokens": 1000000,
      "input_image": true,
      "input_video": true,
      "input_audio": true,
      "reasoning": true,
      "open_weights": true,
      "license": "MIT",
      "price_input_usd_per_mtok": 0.14,
      "price_output_usd_per_mtok": 0.28,
      "price_cache_hit_usd_per_mtok": 0.0028,
      "price_avg_usd_per_mtok": 0.06,
      "free_openrouter": true,
      "intelligence_aa_index": 38,
      "speed_output_tok_per_s": 80,
      "time_to_first_token_s": 2.84,
      "response_500_tokens_s": 9,
      "agent_benchmarks": {
        "swe_bench_verified": 71.0,
        "swe_bench_pro": 56.1,
        "terminal_bench_2_1": 60.7,
        "claweval_pass3_pct": 64.0,
        "gpqa_diamond_rank": "57/130",
        "mmlu_pro_rank": "69/130",
        "livecodebench_rank": "55/135",
        "mmmu_rank": "40/88"
      },
      "benchlm_public_score": 58.3,
      "benchlm_rank": 74,
      "vals_index_pct": 51.6,
      "notes": "Multimodal natif (texte, image, vidéo, audio). Écosystème Xiaomi (MiMo Code, Claw). Verdict rapport : DeepSeek pour texte/codage, MiMo pour multimodal."
    }
  ]
}