Baseten Benchmarks Show Mirendil-Base-Large Beats Rivals on Perplexity

Baseten, now Mirendil's inference partner, independently benchmarked Mirendil-Base-Large on its private codebase using bits per byte, where it beat all tested models including Kimi K3 and DeepSeek, though some users questioned the details.

2026-10-24 ~ 2026-10-24 · 3 related posts

• Baseten: Mirendil-Base-Large posts lowest perplexity on unseen private code — baseten · 2026-10-24
• Baseten's independent eval: Mirendil-Base-Large posts lowest bits per byte of all models tested — shayan_ · 2026-10-24
• Baseten touts Mirendil base model's record-low perplexity; giffmana questions the data — giffmana · 2026-10-24