AMA

Scenario guide

Best AI models for Research / Analyst

A research assistant that reads long PDFs and answers nuanced questions. Weighted toward reasoning, knowledge, long-context, and a saturation-resistant frontier capability score so the ranking stays meaningful as MMLU-style evals saturate.

Rankings use the same scenario weights and cost blending as the interactive leaderboard on AI Model Analyzer. Data is min-max normalised per benchmark; missing scores are skipped without penalty.

  1. 1
    Gemini 3 Pro
    Google
    Score 90.9Q 96.3In $1.25/M
  2. 2
    GPT-5.5
    OpenAI
    Score 87.9Q 93.4In $1.50/M
  3. 3
    Gemini 3 Flash
    Google
    Score 79.8Q 81.2In $0.30/M
  4. 4
    Gemini 2.5 Pro
    Google
    Score 78.5Q 82.5In $1.25/M
  5. 5
    GPT-5.4
    OpenAI
    Score 75.2Q 79.2In $1.50/M
  6. 6
    DeepSeek V3
    DeepSeek
    Score 73.6Q 73.6In $0.27/M
  7. 7
    Gemini 2.0 Flash
    Google
    Score 72.3Q 70.2In $0.10/M
  8. 8
    Claude Opus 4.7
    Anthropic
    Score 72.2Q 80.2In $15.00/M
  9. 9
    Qwen3 235B
    Alibaba (Qwen)
    Score 69.4Q 68.3In $0.20/M
  10. 10
    Claude Opus 4.6
    Anthropic
    Score 69.0Q 76.6In $15.00/M
  11. 11
    Grok 3
    xAI
    Score 68.4Q 72.8In $3.00/M
  12. 12
    o3-mini
    OpenAI
    Score 68.2Q 70.5In $1.10/M
  13. 13
    o1
    OpenAI
    Score 67.1Q 74.5In $15.00/M
  14. 14
    DeepSeek R1
    DeepSeek
    Score 67.1Q 67.9In $0.55/M
  15. 15
    o3
    OpenAI
    Score 66.2Q 72.6In $10.00/M
Open interactive leaderboard Build custom weightsHome