Drex Max debuts: heavily distilled decision model claims top-3 on public benchmarks at $0.06/M tokens
nischay_twt · x · 2026-10-23
A developer released Drex Max, a heavily distilled 'decision model' trained on millions of decision-based samples, claiming top-3 on nearly all public benchmarks evaluated, handling text, images and video frames. Key specs: sub-100ms per decision (under 1s even at 262K tokens), $0.06 per 1M input tokens, and OpenRouter support. The shared benchmark chart shows Drex Max leading decision benchmarks (Decision Index 67.6, JevBench 92.2, DecideBench 98.5) plus agentic scores like OSWorld-Verified 0.780 and BFCL 98.2, beating Jev 1.14 and GPT-6 Luna Decisions at 183ms median latency.
Related event: Drex Max Launches Distilled Decision Model with 100ms Latency(3 posts)→
More from Models
- Physical Agent Capability Index: Astra leads Argon 1349 to 1324 Elo, both fail 4-bottle task — qinzytech · 2026-10-23
- MiniMax M3.1 Flash hits >1K tok/s decode with day-0 SGLang support across NVIDIA and AMD — MiniMax_AI · 2026-10-23
- ChatGPT hit 1B weekly users in ~44 months; Facebook took 103 — FinanceYF5 · 2026-10-23
- Local LLM shootout: Flint 40B MoE vs Qwen3.8-27B on identical RTX A6000s — RichardSocher · 2026-10-23
- MiniMax Quietly Lists New M3.1-Flash Model on Hugging Face — mrtime777 · 2026-10-23
- TokenSpeed ships Day-0 support for MiniMax M3.1 Flash with ~3× faster inference — zhyncs42 · 2026-10-23