Graphs
- One dot per model, coloured by provider; open weight models share a colour
- Higher: better score; further left: cheaper or faster
- Select a dot to see that model below
Score against cost
Score against time
- Failed at least one folio
Not plotted: Gemini 3.5 Flash (high) (no cost recorded; time graph only)
Gemini 4 Argon (multi-step)
Google, low-thinking draft, medium-thinking checks, position 1
- Mean score
- 98.1
- Colour images
- 98
- Microfilm
- 99
- Folios run
- 7 of 7
- Cost
- $0.28 in total, $0.040 a folio
- Time
- 4.5 minutes in total, 38 seconds a folio
Quick answers
- Best score: , 98.1, $0.28, 4.5 min
- Within a point of the best: , 97.3 for $0.062, 4.5× cheaper and 2.4× faster
- Best from Anthropic: , 94.6, $0.74
- Best from OpenAI: , 78.2, $0.75
- Best open weight model: , 70.9, $0.14
- High thinking beats low in 4 of 13 models, at a median 4.2× the cost
- Best on microfilm: , 99