header-langage
简体中文
繁體中文
English
Tiếng Việt
한국어
日本語
ภาษาไทย
Türkçe
Scan to Download the APP
Partnered With:

Google Neon takes the world's top spot in coding agents: defeating Fable 5.2 with lower task costs

Beating AI News Flash: Google Neon Takes World No. 1 in Coding Agent: Beats Fable 5.2 with Lower Task Costs


Google's newly released Gemini 4 Neon has claimed the top spot on the Artificial Analysis Coding Agent leaderboard, surpassing Claude Fable 5.2, which was only released on November 9.


Neon, paired with Antigravity CLI, scored 58.6 points at the highest reasoning tier (rounded to 59 on the leaderboard), ahead of Fable 5.2 and Opus 5.5 with Claude Code, both at approximately 58 points.


This comprehensive evaluation includes three tests — FrontierCode-AA 2.0, Terminal-Bench 4.1, and SWE-Atlas-QnA-AA — each accounting for one-third of the total score. Since it tests the combination of model and Agent tool, rankings are also influenced by the tools themselves.


Based on API usage estimates by Artificial Analysis, Neon costs an average of $6.86 per coding task, compared to $15.47 for Fable 5.2 and $12.61 for Opus 5.5.


Additionally, in the CursorBench 4.1 test, Neon achieved a score of 63.7% without even enabling the highest reasoning tier, surpassing Fable 5.2's 61.3% at its highest tier, at an average cost of only $6.54 per task versus $48.11 for the latter. Even with Neon at its highest tier, the score rises further to 65.4% with a task cost of just $10.67 — less than a quarter of Fable 5.2's.


Gemini 4 Neon has redefined the performance-cost "Pareto frontier" for coding models.

举报 Correction/Report
Correction/Report
Submit
Add Library
Visible to myself only
Public
Save
Choose Library
Add Library
Cancel
Finish