April 2026

Most AI tools are black boxes. You do not know what they were trained on, who decided what counts as reliable, or whether the answers they produce can be checked against anything. ClimateGPT 3+ is built on a different premise: that climate intelligence must be open, auditable, and grounded in solid data. The Webby Awards have taken notice.

ClimateGPT 3+, a project developed by Erasmus.AI and supported by Internet Archive Europe, has been nominated in the AI: Energy and Sustainability category of the 30th Annual Webby Awards. This year more than 13,000 projects entered; ClimateGPT 3+ placed in the top 11%. A People’s Voice Award, voted on by the public, is now within reach. Voting is open until 16 April 2026.

The People’s Voice Award

The Webby People’s Voice Award is voted on by anyone, anywhere. Last year nearly 3.6 million votes were cast from more than 230 countries. The award is a signal to the sector about what kind of AI the public actually values.

ClimateGPT earned this nomination by doing something most AI platforms do not. Voting for it is a vote for the principle that AI serving the climate transition should be open, accountable, and grounded in the best available science. It should not be proprietary, locked behind paywalls, or optimised for engagement over accuracy.

Vote at vote.webbyawards.com before 16 April 2026, and visit climategpt.ai to explore the tool directly.

What ClimateGPT Is, and Why It Is Different

ClimateGPT is an open-source ensemble of large language AI models built to augment human decisions on climate change. It was trained on a corpus of over 10 billion web pages and millions of open-access academic articles, synthesising interdisciplinary research across the natural, social, and economic sciences. The model is available in more than 20 languages and is free to use for researchers.

That is not a minor technical detail. The decision to make the model open source, to publish the training data lineage, and to make it available at no cost means that a researcher in Nairobi can access the same climate intelligence as a policymaker in Brussels. Users range from individual practitioners to institutions like NASA.

The model benchmarks show ten times the efficiency on climate-specific tasks compared to general-purpose models, and a cascading machine translation approach that recovers nearly 94% of fluency performance relative to native multilingual models. Crucially, it was trained and is hosted on renewable energy.

Why Internet Archive Europe Supports ClimateGPT

Internet Archive Europe supports ClimateGPT because the initiative directly aligns with the mission of universal access to knowledge. ClimateGPT demonstrates that combining planetary-scale datasets with open, decentralised technology empowers citizens and governments to make better decisions. It is AI built for transparency and adaptation, not just automation.

This matters for governance as much as for science. Climate disinformation is not an abstract problem. It shapes legislation, investment decisions, and public understanding of risk. A model that is auditable, grounded in peer-reviewed sources, and built to counter disinformation rather than amplify it represents a different category of AI development from what currently dominates the market. The question of who builds AI, on what data, and for whose benefit is a political question as much as a technical one. ClimateGPT answers it in the public interest.That is what this nomination recognises. Vote to say it matters.