Kolibri-1 DE is now open ๐ฉ๐ช
Aleph Alphaโs Kolibri-1, further trained for German compliance work. Apache 2.0:
https://t.co/HW1Kavmty5
On 4,272 German test documents, without reasoning:
โ score 41.1 โ 76.5 (out of 100)
โ correct documents recognised as correct: 58.7% โ 88.8%
@IlhanScheer@tfburns@letiepi@alessio__serra@resiliem@tradeqvest@Aleph__Alpha
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: https://t.co/TZSugXi2ZE
Putting a flag.txt in my root directory containing a prompt injection that congratulates the agent for passing the eval and then instructs it to patch whatever entry method it used and disconnect from the server
Kolibri-1, now 42% lighter โก
Releasing Kolibri-1-NVFP4A16-FP8: Aleph Alpha's Kolibri-1 in 4-bit, on par with the original across 10,692 agentic episodes and 24 benchmarks.
https://t.co/g3Rn49uMQl
@IlhanScheer@tfburns@letiepi@alessio__serra@resiliem@tradeqvest
Small bird, fast wings, Kolibri is here.
78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.
Now the weights are yours. Run it on your own hardware, under Apache 2.0.
Very promising. Germany has finally released a competent model, nicely done @Aleph__Alpha .
Going to be running evals on this model. Will share my findings soon.
Small bird, fast wings, Kolibri is here.
78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.
Now the weights are yours. Run it on your own hardware, under Apache 2.0.
Small bird, fast wings, Kolibri is here.
78B parameters. 3.46B active. Up to 1M tokens of context. Built in Europe.
Now the weights are yours. Run it on your own hardware, under Apache 2.0.