We present the Consciousness Gradient Index (CGI), a universal 0–10 scale measuring integration capacity across living and collective systems. A monkey’s behaviour is wildly different from a cell’s behaviour. But the mechanisms that let each system act as a whole can be compared. A single cell coordinates its internal chemistry; a monkey coordinates billions of neurons; an ant colony coordinates thousands of individuals. Different substrates, different scales-but all are systems holding themselves together as a coherent whole. An ant colony with 10⁶ agents yields CGI ~2.2, rivalling a simple nerve net; a human brain yields CGI = 10. You cannot compare what systems do, but you can compare how well their parts work together. That is what CGI measures. The framework defines three regimes: pre-neural (0–2), where integration occurs via cellular signalling; collective (0–3), where integration emerges through distributed coordination; and neural (2–10), where integration scales with synaptic connectivity. Within the neural regime, four threshold bands mark qualitative transitions in cognitive capacity, validated against 11 species spanning eight orders of magnitude in neuron count. Phase transition analysis reveals that collectives show sharp transitions in connectivity (at the percolation threshold) but smooth saturation in agent count and gradual degradation with noise. This connectivity-driven phase transition independently recovers phenomenology documented in two decades of swarm criticality research, where starling flocks and fish schools operate near critical points with approximately 6–7 topological neighbours per agent. The mechanistic basis of regime boundaries lies in network architecture: collectives are sparse, flat, and slow; neural systems are dense, hierarchical, and plastic. The transition from “dim” collective integration to higher neural regimes requires not just more elements, but qualitative architectural innovations that collectives fundamentally lack.