LMArena launches six new agent indexes; Claude Fable 5.2 tops Work, Coding and Chat
arena · x · 2026-11-19
LMArena introduced six new focused indexes built from millions of real agentic sessions, each answering a specific model-selection question: Work, Coding, Chat (using all four agent performance signals), Task Outcome (task completion + praise vs. complaint), Tool Use (Bash reliability), and Steering (steerability).
Screenshots (Nov 13, 65 models) show Claude Fable 5.2 (Max) ranking #1 on the Work Index (1452), Coding Index (1533), and Chat Index (1386), with Claude Opus 5.5 second. The Daily Driver tier is led by GPT-6.1 Terra (Max) and Gemini 4 Argon (High) at roughly 1/10th the price of frontier models, while GPT-6 Luna (Max) anchors the Efficient tier at 1/100th the price. The #1 Work Index model costs $11.36 per task at 25.1 minutes, versus $1.59 for third-ranked Gemini 4 Carbon. More signals will be added over time.
More from Models
- Abacus AI open-sources uncensored Smaug, a 27B model aimed at agentic work — bindureddy · 2026-11-19
- GAO audit finds Claude, Grok, ChatGPT and Gemini safeguards easily bypassed — midwoodgirl10 · 2026-11-19
- Harmonic's Aristotle rewrites Bitcoin Core's C signature parser in Rust with a formal proof — satnam6502 · 2026-11-19
- GPT-6.2 Astra users note it can work 30 minutes without any progress updates — Angaisb_ · 2026-11-19
- Arena points readers to the live Agent Arena Index and its methodology — arena · 2026-11-19
- Investor complains of 'rude' Anthropic rate limit, blames Nvidia deal monopoly — StewartalsopIII · 2026-11-19