Executive summary
The results show that a broad CPS library can support accessible foreground choices, but a palette category can still be visually cohesive while containing few internal text pairs. Pastel and bright collections require especially deliberate text and surface assignment.
CPS reference findings
- 648 references (46.8%) meet 4.5:1 against white.
- 699 references (50.4%) meet 4.5:1 against near-black.
- 1,037 references (74.8%) reach 7:1 with at least one of those text candidates.
- Near-black is the preferred text candidate for 719 references; white is preferred for 667 references.
Palette-template findings
Each five-color palette contains ten unique unordered pairs.
- 316 of 450 palettes contain at least one pair at or above 4.5:1.
- 273 palettes contain at least one pair at or above 7:1.
- 371 palettes contain at least one pair at or above 3:1.
- The average palette contains 2.28 AA-capable pairs.
Results by palette category
| Category | Palettes with AA pair | Palettes with AAA pair | Average max contrast |
|---|---|---|---|
| RGB | 47/50 (94%) | 45/50 (90%) | 10.25:1 |
| CMYK | 49/50 (98%) | 48/50 (96%) | 10.91:1 |
| Greyscale | 47/50 (94%) | 45/50 (90%) | 14.71:1 |
| HTML Color | 50/50 (100%) | 48/50 (96%) | 13.43:1 |
| Printing | 38/50 (76%) | 27/50 (54%) | 8:1 |
| Gradient | 18/50 (36%) | 11/50 (22%) | 4.85:1 |
| Pastel | 1/50 (2%) | 1/50 (2%) | 1.49:1 |
| Dark | 48/50 (96%) | 45/50 (90%) | 11.41:1 |
| Bright | 18/50 (36%) | 3/50 (6%) | 4.86:1 |
Pastel palettes had the lowest internal contrast coverage in this dataset. Greyscale, HTML Color, Dark, CMYK, and RGB categories produced the strongest average maximum contrast.
Method
The report analyzes the CPS v2.0.0 dataset and the Color Pick palette catalog available on the publication date. Contrast ratios are calculated from opaque sRGB values using relative luminance. Every unordered pair within each five-color palette is tested at 3:1, 4.5:1, and 7:1 thresholds.
The CPS portion tests every reference against #FFFFFF and #111111. Full generation, conversion, uniqueness, and release-validation details are documented in the CPS methodology.
Primary accessibility source: W3C Understanding Contrast (Minimum).
Limitations
- Contrast results evaluate opaque sRGB colors without typography, opacity, images, state changes, or surrounding context.
- A passing pair is not a complete accessibility certification.
- CMYK and total-ink values are planning approximations without a printer-specific ICC profile.
- Color-vision accessibility requires redundant cues and user testing beyond numerical contrast.
Category comparisons describe this specific curated catalog. They should not be generalized to every pastel, bright, dark, print, or web palette.
Suggested citation
Color Pick. “CPS Accessibility Benchmark 2026.” Version 1.0.0, July 16, 2026. https://colorpick.site/research/cps-accessibility-benchmark-2026/Machine-readable report data is available at /data/cps-accessibility-benchmark-2026.json.