Kolibri by Aleph Alpha

“That's the beauty of Apache 2.0! People can just serve it :)”

“that is exactly the capability you need for these kinds of use-cases”

“Awesome work. The work truly needs it!”

“We believe this size and sparsity allows achieving great inference throughput at reasonable levels of intelligence.”

“Good initiative on the sovereignty front though.”

“This was really pleasant to read — especially so for the presentation of a new model.”

“the depth of the tech report shows that they a serious about openness”

“With Schwarz, Cohere, SAP, NVIDIA and the like as backers society can’t have souvereignty in any meaningful way.”

“It can't seem to think things through by itself and then make more tool calls to fetch more information.”

“I sure hope no one actually uses this as part of any aerospace engineering”

“Kolibri performs very well in our evaluation”

“78B MoE with A3.6B is a very nice size.”

“It'll catchup within a year.”

“It's the first time I see this level of openness.”

“We spent considerable time and effort to go beyond just translating.”

“They were very aware of the dangers of such an approach.”

“It's been surprisingly good at the things I threw at it”

“it's actually a good model that also works well on coding and agentic tasks”

“it is trained to say "I don't know" when the answer isn't in the context.”

“We're just happy to share a European alternative in the small model space”

“Qwen 3.8 27B beats Kolibri on nearly every single benchmark.”

“they only have fp16 which won't fit on my Mac”

“This alone makes it much more valuable than many high-profile releases”

“This thing is worse than a Qwen3.8 27B.”

“Qwen3.8 27B beats Kolibri 79.9 vs 70.8 in German in Kolibri's harness on Kolibri's benchmark.”

“I think it’s a safe assumption that they’re leaning into “sovereign” because performance is bad.”

“This model isn't terrible , at least on the benchmarks.”

“Thank you Aleph Alpha team for making it open.”

“Im surprised by how well this works.”

“The absence of any comparison to Qwen3.8 Flash, another MoE model with a small-ish (6B) number of active parameters, is pretty striking.”

“not every use-case is coding a SV-style startup app. That market is quite saturated, so it would make sense to create something locally”

“good at reading German documents and reasoning over them”

“Was expecting that a "sovereign" AI model would at least use their own sovereign language (German) on the website as one of the options.”

“The benchmarks are impressive given the problem space they are working in.”

“being trained to say "I don't know" instead of misleadingly talking confidently about what it does not know is interesting in itself”

“Interesting they recommended high end software without considering quant 4 or 8”

“spends way too many tokens on overthinking stuff”

“You can use Kiwix, OpenStreetMap and Kolibri as an AI-free equivalent.”

“we do a lot of work to make an offline first education platform”

“designed to be deployed in communities that lack consistent access to the public internet”