Individual experiment result
GermEval NER (DE)
Kolibri-1 (FP8) Aleph Alpha
Micro-F1 excluding MISC reasoning off fp8
72.7%
Micro-F1 excluding MISC
Run details
- Date
- 2026-10-05
- Test cases
- 1,024
- Median latency
- 0.33s
- Total cost
- —
- Avg. prompt tokens
- 1,059.7
- Avg. answer tokens
- 24.9
- Avg. reasoning tokens
- —
- Quantization
- fp8
About this benchmark
Native German named-entity recognition (person/location/org/misc) emitted as JSON; seqeval micro-F1 excluding MISC.
Provenance
- Canonical model
- aleph-alpha/kolibri-1-fp8
- Run ID
2026-10-05T17-00-36__openai__Aleph-Alpha__Kolibri-1__germeval_de