Best value on MathVista

“Value” is normalized benchmark score (0–100 for this leaderboard cohort) divided by input price per million tokens. Higher means more capability per dollar on this axis only — always sanity-check latency, context length, and your real workload.

1. 1

   Gemini 2.0 Flash

   Google
   622.60

   62.3 / $0.10/M
2. 2

   Llama 4 Scout

   Meta
   309.17

   55.6 / $0.18/M
3. 3

   Gemini 1.5 Flash

   Google
   290.13

   21.8 / $0.08/M
4. 4

   Llama 4 Maverick

   Meta
   236.70

   63.9 / $0.27/M
5. 5

   GPT-4o mini

   OpenAI
   113.87

   17.1 / $0.15/M
6. 6

   o3-mini

   OpenAI
   76.38

   84.0 / $1.10/M
7. 7

   Gemini 2.5 Pro

   Google
   76.03

   95.0 / $1.25/M
8. 8

   Gemini 1.5 Pro

   Google
   29.53

   36.9 / $1.25/M
9. 9

   Grok 2

   xAI
   25.48

   51.0 / $2.00/M
10. 10

    Claude Sonnet 4

    Anthropic
    25.25

    75.8 / $3.00/M
11. 11

    Grok 3

    xAI
    22.50

    67.5 / $3.00/M
12. 12

    o1-mini

    OpenAI
    17.91

    53.7 / $3.00/M
13. 13

    Claude 3.5 Sonnet

    Anthropic
    15.79

    47.4 / $3.00/M
14. 14

    GPT-4o

    OpenAI
    14.66

    36.6 / $2.50/M
15. 15

    o3

    OpenAI
    10.00

    100.0 / $10.00/M
16. 16

    Claude Opus 4

    Anthropic
    6.15

    92.3 / $15.00/M
17. 17

    o1

    OpenAI
    4.30

    64.5 / $15.00/M
18. 18

    Claude 3 Opus

    Anthropic
    0.00

    0.0 / $15.00/M

AI Model Analyzer does not recommend specific vendors; rankings are derived from public data only.