ResearchSep 24, 2026 · 23:402 sources

CLM-8B Hits 87.6% on Terminal-Bench to Set New Agentic Coding Record

From marketbrief.now

Image: marketbrief.now
AI;DR

CLM-8B, a Contrastive Language Model, achieved 87.6% on Terminal-Bench and 81.6% on the DeepSWE agentic coding benchmark, delivering up to 9x faster inference than the Jev model. The model uses a contrastive learning objective to connect states and actions, and its developers reduced inference latency by disaggregating states and actions so their embeddings can be cached and reused independently.

Read the sourceSign in to suggest

Sources

#clm-8b#terminal-bench#agentic-coding#inference#benchmark#llm#coding#reasoningMarkdown
Content history (2)
  • 2026-10-04 · Summary · vi · Replaced stored text
    Mô hình CLM-8B vừa đạt điểm 87.6% trên bảng chuẩn đoán Terminal-Bench, vượt mặt các đối t…
    Mô hình Contrastive Language Model mới mang lại tốc độ suy luận nhanh hơn tới 9 lần so vớ…
  • 2026-10-04 · Title · vi · Replaced stored text
    CLM-8B đạt 87.6% trên Terminal-Bench, lập kỷ lục mới cho lập trình agent
    CLM-8B đạt 87,6% trên Terminal-Bench và thiết lập kỷ lục lập trình agentic mới

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today