COST Open Call 2026 (OC-2026-1): Up to 80 New Pan-European Research Networks Funded at Roughly €600,000 Each, Closing 28 October 2026

COST Open Call 2026 (OC-2026-1): Up to 80 New Pan-European Research Networks Funded at Roughly €600,000 Each, Closing 28 October 2026

There is a category of research problem that no single laboratory can solve and no single grant can address, because the bottleneck is not money for experiments — it is that the twenty people across Europe who hold the relevant pieces have never been in a room together. COST exists for that category. The COST Association funds four-year science and technology networks called COST Actions, and its 2026 Open Call, formally collection OC-2026-1, opened on 31 July 2026 at 12:00 noon CEST and closes on 28 October 2026 at 12:00 noon CET.

This cycle funds up to 80 new Actions, an increase from the 70 funded in the previous call. Each approved Action receives on average approximately €600,000 across its four-year lifetime — COST’s own estimate is around €140,000 in the first year and on average €180,000 per year for the remaining three. Crucially, that money does not fund research. It funds the meetings, training schools, mobility grants, and dissemination that let nationally funded research groups coordinate across borders.

Understanding that distinction is the single most decisive factor in whether a COST proposal succeeds. Reviewers are explicitly asked to judge whether a pan-European network — rather than a research project — is the right instrument for the challenge you describe. Proposals that read like a rejected Horizon Europe project with the budget removed do not get funded.

This guide is built from the official OC-2026-1 Open Call Announcement (published 31 March 2026), the COST Open Call Proposers’ Guidelines COST 104/21 V5.0, and COST’s own published notes on what changed for this cycle.

Key Details at a Glance

What COST Actually Funds — and What It Does Not

COST has been running since 1971 and is funded through Horizon Europe, but it is an intergovernmental framework rather than an EU instrument, a distinction the guidelines ask you to observe even in your writing style. Its mission is to strengthen Europe’s capacity to address scientific, technological and societal challenges by funding bottom-up, excellence-driven, open and inclusive networks in all areas of science and technology.

The Proposers’ Guidelines are blunt about the boundary: “The research and development activities needed for the achievement of the Action objectives are not funded by COST and rely on nationally or otherwise funded research projects and resources (e.g., employees’ time, infrastructures and equipment).”

What the grant does cover:

• Meetings — Management Committee meetings and Working Group meetings.
• Training Schools — structured teaching events for the network and beyond.
• Mobility of researchers and innovators — Short-Term Scientific Missions (STSMs) and Virtual Mobility.
• Conference grants — ITC Conference Grants, Young Researcher and Innovator Conference Grants, and Dissemination Conference Grants for presenting at third-party conferences.
• Dissemination and communication products.
• Expenses incurred for the benefit of the network.

Funding flows through yearly Action Grant Agreements based on annual Work and Budget Plans decided by the Action’s Management Committee. One consequence matters at the proposal stage: no budget forecast is requested when submitting. You do not build a costed workplan. You describe activities, and the COST Association allocates budget to approved Actions based on its own parameters and available funds. Applicants used to writing detailed justification-of-resources sections should resist the urge to smuggle one in — the space is better spent on the network rationale.

Who Can Propose, and the Two Hard Network Requirements

COST Actions are open to anyone with a legal affiliation in a COST Member, a Near Neighbour Country, or a Third State. Eligible legal entities include public authorities at any level, universities, research centres, SMEs, large companies, NGOs, and any other legal entity recognised under a national or international framework.

For the network of proposers specifically, two quantitative rules govern eligibility, and both are checked mechanically:

Rule one — country coverage. The network must include proposers affiliated to legal entities from at least seven COST Full or Cooperating Members, of which at least 50% must be COST Inclusiveness Target Countries (ITC). The guidelines publish a table mapping the requirement upward: seven or eight member countries requires four ITCs, nine or ten requires five, eleven or twelve requires six, and so on up to forty-two countries requiring twenty-one ITCs.

The ITC list currently comprises fifteen EU Member States — Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Greece, Hungary, Latvia, Lithuania, Malta, Poland, Portugal, Romania, Slovakia and Slovenia — plus specified EU outermost regions (French Guiana, Guadeloupe, Martinique, Mayotte, Reunion Island and Saint-Martin for France; Azores and Madeira for Portugal; the Canary Islands for Spain), and ten Full Members that are not EU Member States: Albania, Armenia, Bosnia and Herzegovina, Georgia, Moldova, Montenegro, North Macedonia, Serbia, Türkiye and Ukraine. Note the outermost-region rule for proposer counting: a proposer in the Azores counts as Portugal (ITC), while one in Guadeloupe counts as France (non-ITC).

Rule two — Young Researchers and Innovators. This is the headline change for 2026. At least 40% of the individuals in the network must be Young Researchers and Innovators — defined in the COST Glossary as a researcher or innovator under the age of 40 — during the calendar year of the Collection Date. This is a new eligibility criterion layered on top of the long-standing country rules, and it will disqualify networks assembled purely from established professors.

Also new for this cycle: an individual may participate as either Main or Secondary Proposer in one proposal only per collection. The platform blocks multiple participation, and bypassing the rule has escalating consequences — a Main Proposer who submits more than one proposal has all of them declared ineligible.

No letters of intention are required from any institution involved. Every proposer needs a registered, up-to-date e-COST profile, and only one profile per person.

What Goes Into a Proposal

A COST Action proposal has four parts, all completed inside e-COST:

General Features. A title of at most 12 words. An original acronym not already used by any other public or private entity or research group, containing only letters, numbers, hyphens and the “@” symbol. A summary of at most 250 words, which is shown to potential Independent External Experts when their availability is checked and which is published on the COST website if the Action is approved. Between one and five key expertise entries selected from a drop-down across six S&T fields. Three to five keywords, each under 60 characters.

That keyword instruction deserves attention: COST explicitly warns against generic keywords such as “interdisciplinary”, “research coordination”, “science” or “networking”, because keywords drive expert matching. Vague keywords produce mismatched reviewers, and mismatched reviewers produce mediocre scores.

Technical Annex. Capped at fifteen pages, roughly equivalent to 45,000 characters, and — new this cycle — completed directly in e-COST rather than uploaded as a PDF. The platform now refuses submissions that exceed page or text limits, which removes a historically common source of ineligibility. It is structured around the three evaluation criteria:

• Excellence in S&T and Networking: main challenge; objectives; state-of-the-art; rationale for choosing networking; critical mass of the network.
• Impact: impact related to objectives; involvement of stakeholders; communication, dissemination and valorisation.
• Implementation: Action structure; work plan; deliverables; and a mandatory Gantt chart.

References. Not mandatory but recommended, capped at 500 words.

Network of Proposers. The Main Proposer enters each Secondary Proposer’s first name, last name and email — and the email must exactly match the one on that person’s e-COST profile, or the invitation will not resolve. Experts never see individual names. They see aggregated features only: ITC percentage, number of proposers, geographic distribution, gender distribution, YRI percentage, expertise distribution, and institutional distribution.

Anonymity: The Rule That Disqualifies Good Proposals

COST runs a double-blind evaluation. The Technical Annex must contain neither direct nor indirect reference to the proposers or their institutions. Proposals can be declared ineligible at any point in the process for breaching this.

The guidelines give worked examples of fatal statements, and they are instructive precisely because several look harmless:

• “Prof. Smith will coordinate the Action activities within WG5” — direct reference.
• “Several members of the proposers’ network have been involved in previous FP7 projects, like XYZ, and institutions/organisations/networks/COST Actions, such as CAXXXXX” — indirect reference, and note that it is fatal irrespective of the size of the project mentioned.
• “Among government-run public services we have the Department of Health of XXX on board” — direct reference.
• A claim that the network “has already generated some output” with a footnote linking to a video or webpage where Secondary Proposers are identifiable — indirect, potentially identifiable.

Two related constraints: the proposal must contain no links or references to additional information beyond the Technical Annex — no project websites, no audio-visual material. And in the References part, you may cite proposers’ own publications, provided nothing in the text (bold, italics, links) signals that a proposer authored them.

The practical discipline here is to write the proposal anonymously from the first draft rather than trying to scrub identity out at the end. Retrofitted anonymisation is where the indirect references survive.

COST also publishes an explicit position on generative AI: proposers must be human beings; they are fully and solely responsible for all submitted content regardless of how it was authored; proposals are judged solely on content; and proposers are warned that such tools may supply ideas that are not new, suggest existing text fragments, or generate content that could be read as scientific misconduct.

How Proposals Are Scored

Three Independent External Experts carry out a remote, double-blind peer review, each submitting an Individual Evaluation Report. Experts are matched using the research areas and keywords you supplied.

The scoring is asymmetric and worth internalising:

Total range: 0–13. Scores on the first two criteria run empty section (0), poor (1), fair (2), good (3), very good (4), excellent (5). Implementation uses only four values: empty section (0), insufficient (1), standard (2), outstanding (3).

Proposals failing to reach the threshold on either Excellence or Impact are not selected for funding, regardless of the total. Implementation carries no threshold and a smaller range — it is a differentiator, not a gate. A proposal with a brilliant workplan and a merely “fair” Impact section is dead; the reverse is survivable.

After individual reports, one expert is appointed Rapporteur to coordinate and draft a Consensus Evaluation Report. Consensus is sought but not imposed. The COST Administration quality-checks every report, and a Resolution Committee — chaired by the Head of Science Operations, with Open Call scientific management, the relevant Science Officers, external experts and, where needed, a Legal Officer — resolves divergences and inconsistencies, particularly where a mark would change.

The Administration then builds a shortlist by score, excluding anything below threshold on Excellence or Impact. The Scientific Committee assesses acceptability, endorses the shortlist, and the Committee of Senior Officials (CSO) takes the final funding decision.

Ties at the cut-off are broken sequentially: excellence mark, then impact mark, then implementation mark, and only then by an automated random selection among still-tied proposals. That tie-break rule is a quiet argument for pushing the Excellence section as hard as possible — it is both a threshold and the first tiebreaker.

Timeline and What Happens After Submission

A few mechanics of the submission window matter. A proposal has “draft” status until submitted, and can be revised as many times as needed until the Collection Date — but each revision reverts it to draft, and it must be submitted again to be evaluated. Only the last submitted version counts. Only proposals in “submitted” status are evaluated at all.

COST recommends against submitting immediately before the deadline and states plainly that it cannot be held responsible for delays caused by heavy traffic or connection problems. Data from unsubmitted proposals is deleted after the Collection Date, and submitted proposals become read-only — nothing is transferable to a later collection, so a resubmission next year means rebuilding in the new template.

The Technical Annex of an approved proposal becomes the basis of the Action’s Memorandum of Understanding, which is a reason to write it as a document you would be content to be held to for four years, not as a sales pitch.

Preparation Strategy for the Remaining Weeks

With the deadline on 28 October 2026, the constraint for most teams is not writing — it is assembling a compliant network. Work in this order.

Fix the arithmetic first. Before drafting anything, list your candidate proposers, mark each one’s country against the ITC table and each one’s YRI status, and check both ratios. If you have nine member countries you need five ITCs; if you are at 35% YRI you need to recruit, not reword. These numbers are computed by the platform and are not negotiable.

Contact your COST National Coordinator early. The announcement and guidelines both point proposers to their CNC for guidance, and the list is at cost.eu/cnc. CNCs also nominate Management Committee members if the Action is approved, so the relationship has downstream value. Secondary Proposers affiliated in a Full or Cooperating Member can flag their interest in MC nomination when accepting the invitation.

Write the networking rationale before the science. Criterion 1D asks you to demonstrate why a pan-European network — as opposed to a research project — is the appropriate instrument. If you cannot answer that in a paragraph without hedging, the proposal is not yet a COST proposal.

Write objectives in both registers. The guidelines require objectives spanning research coordination (common definitions, coordinated methodologies, coordinated experimentation or testing, comparative data analysis) and capacity building (knowledge exchange, bridging disciplines, enabling collaboration among academia, industry, policymakers and NGOs). Proposals with only one register read as incomplete against the criterion.

Separate deliverables from activities. COST states directly that networking activities — meetings, training schools — are not deliverables. Deliverables are tangible outputs: reports, documents, technical diagrams, papers, publications, training-school content, input to standards, best practices, white papers.

Respect the style guide. COST asks for capitalised Action-specific terms (COST Action, Working Group, Short-Term Scientific Mission), all acronyms explained, “Europe” or “COST Member Countries” rather than “EU” for geographical scope, “framework” or “scheme” rather than “EU instrument”, no first-person pronouns, and “the Action will” rather than “the proposed Action is planned to”. These are small signals, but reviewers read dozens of proposals and register register.

Common Mistakes

Addressing COST policy in the wrong section. The guidelines state that country coverage, gender balance and involvement of Young Researchers and Innovators must not be addressed in the critical-mass criterion. Those figures reach reviewers automatically through the aggregated network features. Spending Criterion 1E on them wastes space and ignores an explicit instruction.

Treating anonymity as a formatting pass. Covered above, and worth repeating because it produces ineligibility rather than a low score — there is no partial credit.

Building a budget. No budget breakdown is requested. Allocation happens after approval.

Submitting then editing without resubmitting. The revert-to-draft behaviour catches people every cycle.

Contacting anyone about your proposal. The guidelines are unambiguous: proposers must not, under any circumstances, contact actors involved in evaluation, selection or approval about their proposal. Any attempt may lead to immediate exclusion.

Duplicating your own proposal. A submitted proposal must not be identical, even in part, to another in the same collection; if it happens, only the first-submitted is considered.

Frequently Asked Questions

Do I need to be at a university? No. Public entities, research centres, SMEs, large companies and NGOs all qualify, as does any other recognised legal entity. Individuals with no legal affiliation cannot be proposers.

Can researchers outside Europe participate? COST Actions are open to anyone with a legal affiliation in a COST Member, a Near Neighbour Country, or a Third State (International Partner Country), and participation from Near Neighbour Countries is described as particularly welcome. Eligibility to serve as Main versus Secondary Proposer varies by affiliation type, so check Table 1 of the Proposers’ Guidelines before assigning roles.

What exactly is a “Young Researcher and Innovator”? The COST Glossary defines it as a researcher or innovator under the age of 40, assessed during the calendar year of the Collection Date. It is an age criterion, not a career-stage one.

How competitive is it? COST has not published a success rate for this collection, but the increase from 70 to 80 funded Actions is described by COST itself as improving the success rate. Historically the call attracts many hundreds of proposals; treat it as genuinely competitive and do not rely on threshold-clearing scores alone.

What if I disagree with the outcome? A redress procedure exists. The Main Proposer may submit a request to [email protected] within 15 calendar days of being notified of ineligibility or of the evaluation result, citing the proposal number and title and describing the alleged procedural shortcoming or factual error. Redress is limited to procedure and factual error — challenges to the experts’ expert judgement are inadmissible, and acceptability decisions are not subject to redress at all. The outcome is definitive.

Will I see the reviews? Yes. Consensus Evaluation Reports are made available to all proposers via e-COST.

Can I reuse an unsuccessful proposal next year? You will need to rebuild it. Submitted proposals become read-only after the Collection Date and no section transfers to another collection.

Official Links and Next Steps

• Open Call page: cost.eu — Open Call: A Simple One-Step Application Process
• Documents and Guidelines, including the OC-2026-1 Announcement and the Proposers’ Guidelines (COST 104/21 V5.0): cost.eu/funding/documents-guidelines
• Submission platform: e-COST at e-services.cost.eu
• COST National Coordinators: cost.eu/cnc
• Enquiries: [email protected], or the “contact us” link inside e-COST
• Redress requests only: [email protected]

If you are considering this call, the productive next step is not drafting — it is a spreadsheet. Name the countries, mark the ITCs, count the under-40s, and confirm that no one on your list is already committed as a proposer elsewhere in this collection. Once those four columns are green, you have a proposal worth spending fifteen anonymous pages on. Until then, you have a good idea without an eligible network, and COST will not evaluate the idea.