Newsroom automation for Turkish local publishers · bultenhub.com
BültenHub monitors Turkish news sources around the clock, groups stories by event and writes original, source-linked news with its own models. It then turns them into bulletins, voice-overs and news videos.
| Model | What it does |
|---|---|
| BHAI | Qwen3.5-4B LoRA that writes a headline and a news text from a Turkish source. Runs in production on llama.cpp. |
| BHKAI | Fast Turkish news category classifier (TF-IDF + SGD). Retrained nightly, promoted only if it passes its exam. |
| BHDAI | Same-event matcher that groups stories from different sources. Retrained nightly, promoted only if it passes its exam. |
Models are retrained at night and published only after passing their exam. Updated 2026-11-13.
| Model | Version | Date | Exam |
|---|---|---|---|
| BHAI | qwen35-4b-hoca-mini2-20261031 | 2026-10-31 | exam A score 0.852, grounding 0.986, coverage 0.745 |
| BHKAI | bhkai-20261112 | 2026-11-12 | at confidence ≥ 0.8: coverage 48.1% (was 47.8%), agreement with teacher 96.9% (was 96.5%) |
| BHDAI | bhdai-20261112-c03 | 2026-11-12 | at threshold 0.1: same-event pairs passed without a call 69.2% (was 69.7%), hard negatives wrongly passed 3.14% (was 3.22%) |
bhkai-20261112: at confidence ≥ 0.8: coverage 48.1% (was 47.8%), agreement with teacher 96.9% (was 96.5%)bhdai-20261112-c03: at threshold 0.1: same-event pairs passed without a call 69.2% (was 69.7%), hard negatives wrongly passed 3.14% (was 3.22%)bhdai-20261111: at threshold 0.1: same-event pairs passed without a call 69.1% (was 68.9%), hard negatives wrongly passed 2.61% (was 2.61%)bhkai-20261110: at confidence ≥ 0.8: coverage 45.2% (was 44.4%), agreement with teacher 97.4% (was 97.3%)bhdai-20261110-adet120k: at threshold 0.1: same-event pairs passed without a call 67.0% (was 68.5%), hard negatives wrongly passed 2.39% (was 2.60%)bhdai-20261105: at threshold 0.1: same-event pairs passed without a call 67.0% (was 67.9%), hard negatives wrongly passed 2.59% (was 2.73%)bhkai-20261104: at confidence ≥ 0.8: coverage 57.2% (was 57.7%), agreement with teacher 97.4% (was 97.2%)bhdai-20261104: at threshold 0.1: same-event pairs passed without a call 70.0% (was 69.5%), hard negatives wrongly passed 2.69% (was 2.69%)bhkai-20261103: at confidence ≥ 0.8: coverage 46.3% (was 46.3%), agreement with teacher 97.2% (was 95.3%)bhdai-20261103: at threshold 0.1: same-event pairs passed without a call 68.6% (was 68.0%), hard negatives wrongly passed 2.90% (was 2.94%)bhkai-20261101: at confidence ≥ 0.8: coverage 56.1% (was 56.1%), agreement with teacher 96.5% (was 96.7%)qwen35-4b-hoca-mini2-20261031: exam A score 0.852, grounding 0.986, coverage 0.743BHAI demo: paste your own notes and get a Turkish news draft (free ZeroGPU).
🇹🇷 BültenHub, Türkiye'deki yerel yayıncılar için haber otomasyonu platformu. Modellerimiz açık; geri bildirimlerinizi bekliyoruz.