About the System

The voluntary carbon market is fragmented across dozens of registries, standards and marketplaces. Each registry uses its own criteria. Each marketplace presents projects differently. For anyone who needs to compare projects, a buyer, an analyst, a researcher, there is no common measure.

CarbonMeld exists to fill this gap: applying a standardized evaluation pipeline to all projects, based exclusively on public documentation, so that comparisons are possible and consistent.

The pipeline collects public documents from the main registries, Verra, Gold Standard, CDM, CORSIA and others, and metadata from active marketplaces. For each project, an LLM-based analysis system evaluates four dimensions: registry integrity, transparency, claim safety, and documentation quality.

No score is assigned manually. The pipeline is identical for all projects, regardless of who developed them or who sells them.

The full criteria, the weight of each dimension and the calculation rules are described in the Methodology, updated and publicly versioned.

Anyone looking to buy carbon credits faces a fragmented market: projects scattered across dozens of marketplaces, prices that cannot be compared, documentation spread across different registries, and no straightforward way to understand what is available or at what quality level.

CarbonMeld brings it all into a single database: which projects are listed, where, at what price, with what documentation, and with what quality score. For buyers, analysts, and ESG teams who need to find their bearings before committing time and budget, it is the starting point.

OTC operators and private marketplaces can also be included in the database, provided their projects are publicly exposed with visible prices and availability. The pipeline is the same for everyone.

CarbonMeld is not a marketplace. It is an independent analysis tool that works before the negotiation, to understand what is worth exploring further and what is not.

A CarbonMeld score is a methodological output, not a certification.

A high score means that the available documentation is solid across all four dimensions. It does not mean the project is risk-free, nor that CarbonMeld recommends purchasing it.

A low score may reflect poor documentation, identified risks in the evidence base, or simply limited available evidence. It is not automatically a negative judgment on the project: it may indicate that documents are not publicly accessible, not that the project is problematic.

The confidence level is the variable to read alongside the score. It indicates how complete the evidence base is on which the score was calculated. A 7.2 with high confidence and a 7.2 with low confidence are not the same thing: in the second case, the available documentation was partial, and the number should be weighted accordingly.

CarbonMeld has three revenue sources: paid processing of new documents submitted by project owners, advertising banners, B2B consulting. All three are publicly disclosed.

None of these commercial relationships influences scores. The pipeline has no access to payment data. Advertisers have no visibility into the methodology or individual assessments. Consulting clients are subject to the same standardized pipeline as everyone else.

The reason we state this here, and not only in the Editorial Policy, is that in a market exposed to greenwashing the question of independence is not a detail: it is the premise on which everything else rests.

CarbonMeld is a screening and comparison tool, useful in the initial stages of analysis.