Advancing the frontier of artificial intelligence through open research and collaborative innovation.
| # | Project Name | Type | Details | Release Date | Size |
|---|---|---|---|---|---|
| 0 | Cleveris Agent | Agent | An autonomous agent that lives on your server, remembers what it learns, and gets more capable the longer it runs. | 04/25/25 | N/A |
| 1 | Cleveris-7B | Model | Competitive olympiad programming post-trained on Code-7B base. | 04/20/25 | 14 GB |
| 2 | Cleveris Math 1 | Model | Small but mighty 30B SOTA mathematician in partnership with leading universities. | 04/15/25 | 60 GB |
| 3 | Cleveris-4.5-Base-38B | Model | Roughly equivalent performance to Cleveris-4-70B at half the model size; post-trained entirely on the Cleveris network. | 04/10/25 | 76 GB |
| 4 | Cleveris 4 Technical Report | Paper | Comprehensive technical report detailing architecture, training methodology, and evaluation results. | 04/05/25 | N/A |
| 5 | Cleveris-4-Llama-3.1-405B | Model | Frontier hybrid-mode reasoning model based on Llama-3.1-405B. | 04/01/25 | 810 GB |
| 6 | Cleveris-4-Llama-3.1-70B | Model | Smaller hybrid-mode reasoning model with 70B parameters. Shares the same improvements as the 405B variant. | 03/28/25 | 140 GB |
| 7 | Cleveris-4-14B | Model | Small and dense Cleveris variant for local inference. | 03/25/25 | 28 GB |
| 8 | Measuring Thinking Efficiency in Reasoning Models | Research | The Missing Benchmark for evaluating reasoning efficiency in large language models. | 03/20/25 | N/A |
| 9 | Cleveris 3 Dataset | Dataset | Complete dataset used in pretraining of Cleveris 3 models. | 03/15/25 | N/A |
| 10 | Sequential Monte Carlo for LLMs | Research | Taming LLMs with Sequential Monte Carlo methods. | 03/10/25 | N/A |
| 11 | Cleveris Network | Training | Open infrastructure democratizing AI development through distributed training. | 03/05/25 | N/A |
| 12 | Atropos Framework | Framework | Language Model RL Environments for advanced reinforcement learning research. | 03/01/25 | N/A |
Cleveris Research is dedicated to advancing artificial intelligence through open research, collaborative development, and transparent publication of our findings. We believe that AI should be developed in the open, with contributions from researchers worldwide.
Our work spans foundation models, reasoning systems, autonomous agents, and the infrastructure needed to train them at scale. We publish our models, datasets, and research papers to accelerate progress in the field.