Aleph Alpha · Kolibri
Kolibri-1
Aleph Alpha's sovereign open-weight MoE (78B total, 3.46B active), trained on German and English only, with no hosted API. Scores are the vendor's own harness at the highest reasoning mode.
- AA Intelligence
- —
- Price
- Self-host
- Output speed
- —
- Context window
- 1.05M tokens
Coding
- LiveCodeBench#7 of 5785.9%
Leader: Fugu at 92.9%
- SWE-bench Verified#12 of 1866.4%
Leader: Claude Sonnet 5 at 85.2%
- Terminal-Bench 2.x#78 of 10527.7%
Leader: Claude Fable 5.1 at 91.4%
Agents and tool use
- τ²-Bench#11 of 8994.7%
Leader: GLM-5.2 at 99.1%
- BrowseComp#15 of 1629.4%
Leader: GPT-6 Astra at 91.5%
Reasoning and knowledge
- MMLU-Pro#7 of 1580%
Leader: Nemotron 3 Ultra 550B A55B at 86.8%
- GPQA Diamond#62 of 14484.3%
Leader: GPT-6 Astra at 96.1%
- Humanity’s Last Exam#76 of 15321.5%
Leader: Claude Sonnet 5.5 at 64.5%
- AIME 2026#5 of 1596%
Leader: ERNIE 5.1 at 99.6%