Jev beats Gemini 2.5 Flash Lite on classifier eval quality and speed
Malte Ubl ran TypeSafe AI's Jev against an existing classifier eval, where it saturated the eval on quality and was 6x faster than Gemini 2.5 Flash Lite.
Jev 1.13.0 vs GPT-6 Luna: JevBench Benchmark Comparison
A JevBench comparison from Benchmark Heaven shows Jev 1.13.0 scoring 86.6% public and 36.7% sealed, while GPT-6 Luna reaches 99.6% and 95.5%. Luna still ranks #30 on JevBench, with the cost axis ($0.135 vs $0.040 per 1,000 decisions) and gate logic below 50 as key factors.
GPT-6 Luna (medium) gets nearly everything right: 99.6% public, 95.5% sealed. Jev 1.13.0: 86.6% and 36.7%. Luna still ranks #30 on JevBench. One column explains it: $0.135 per 1,000 decisions vs $0.040, Cost axis 36.0. Below 50, the gate applies. https://benchmarkheaven.com/jev-models/v1.4.1?compare=jev-1.13.0,gpt-6-luna…
· 0 likesOpen on X
More like this
Fast row classification in DuckDB with Jev
The author released a DuckDB extension that uses TypeSafe AI's Jev to quickly classify rows in CSV/Parquet files or DuckDB tables, taking about 10 seconds for 1k rows—faster than an LLM and more ergonomic than a classifier.
Classifying 1,018 AI Papers with Jev for $0.08
The author summarized each paper with DeepSeek V4 Flash, then sent the title, summary, and 24 candidate topics to Jev for classification, processing 1,018 AI papers at $0.08 total cost and 256ms median end-to-end latency per paper.
