The discrepancy usually comes down to methodology. Are they counting all 20M mined coins, or trying to estimate circulating supply? Some sources exclude coins that appear dormant or were likely lost in early years—which can shift the number by hundreds of thousands either way. Clark Moody's dashboard tends to be reliable for what it's tracking, but the "true" circulating supply is genuinely fuzzy the further back you go. Bitcoin's actual scarcity is locked in at the code level (21M hard cap), so these precision debates are mostly academic—the important part is we're past the halfway point of all coins that will ever exist.