Since the digitalization of academic publications, subscription to journals had taken the form of access rights to content platforms for users. Despite almost 20 years of open access policies, paywalled papers and subsequently subscriptions are still paramount for the oligopoly of publishing industry in its relations with ever-growing consortia of universities and libraries. Nevertheless, in the last five years, new forms of agreement have been signed, notably including paying open access publishing options. The range of names given to these new agreements underlines their complexity and diversity: offset agreements, read and publish, publish and read, transformative agreements. Whereas previously subscription agreements included confidentiality clauses on their content, the transparency requirements of some consortia have led to their availability. Thanks to this publicity and to the funding of the French Minister for Higher Education, Research and Innovation, we will present the preliminary results of a first systematic comparison of more than 60 available agreements between "big" or "medium" publishers (ACS, Elsevier, OUP, Sage, Springer, Wiley, etc.) and European or North American consortia. Drawing on science and technology studies approaches, we have read and analyzed these hundreds-pages long documents, written or translated in English and will present the following results. 
1/ The diversity of agreement transparency conditions: apart from the fact that more than a hundred contracts of this type signed over the last five years are not available, we will show that traces of the culture of secrecy of the previous period are still present and that the "as transparent as possible" logic is dominant 2/ The morphological variety of agreements: we observe a great variety in the size, structure of central objects and content of annexes in the agreements, even if there is clearly a "Springer Compact model" in a limited number of cases.
3/ Shared objects and devices: definition of eligible authors and articles, perimeter of eligible journals for publication, system of recognition of authors and rights holders, financial payment circuits, monitoring of actual costs and number of papers published through the agreement, definition of editorial independence..
4/ A diversity of options for publication: the publishing part of agreements is clearly an output of long and often difficult publications. As a consequence, from the whole collection of agreements, we observe different choices (hybrid and/or full OA journals, limited or unlimited volumes, capped or unlimited costs, …), some consortia signing for constant ones, while others change from one publisher to the other. We will discuss the limitations of these results and the subsequent stages of our research before concluding by raising two issues: first, the establishment of a terminology regarding the financial mechanisms of these agreements; second, the scope of the transformative nature of such agreements on an individual and collective basis.