Story

On air. The free wire runs 15 minutes behind live. See pricing

World, crypto, finance and tech news, scored for impact.

AITechFlowChinese, translatedFiled 10 h ago

"North American Soybean Bun" Google Finally Kills Its Way Back to the Center of the AI Table

Original headline“北美大豆包”谷歌,终于杀回 AI 牌桌中央

Open the sourceReported by TechFlow, techflowpost.com, 20 November 2026

Not investment advice. Muhabbit scores how much a story may matter, not which way to trade.

Google DeepMind released Gemini 4 Neon on November 18, the fourth model in its Gemini 4 series in seven weeks, positioning it between the heavy Carbon flagship and the lightweight Hydrogen model at $3 per million input tokens and $15 per million output tokens. Neon scored 61 on the AA Intelligence Index v5.0.1 at its xhigh setting, ranking third globally behind GPT-6.2 Astra and Claude Fable 5.2 while beating its own Carbon high setting (60), Claude Opus 5.5 (59) and GPT-6.2 Sol (57), and it topped the AA Coding Agent Index at 58.6. The model's economics are the headline: a $6.44 average cost per AA benchmark task, about 17% below Carbon's $7.76, and roughly $6.86 per coding task versus $12.61 for Opus 5.5 and $15.47 for Fable 5.2, with output speed near 110 tokens per second, double Carbon's rate. It also led on CursorBench 4.1 (65.4%), SWE-sweep in Antigravity (39.0%), Zapier AutomationBench (56.3%), Vending-Bench 2 ($16,843) and RiemannBench (84.0%). Watch whether the pricing anomaly persists—Neon's list price exceeded Carbon's and Argon's promotional rates until both flagships revert to $4/$20 on December 1—and whether xhigh's verbosity (283 million output tokens across the AA suite, 56.5-second first-token latency) pushes enterprise buyers toward the medium tier.

Muhabbit's AI wrote this summary in its own words from TechFlow's report and translated it from Chinese. It can get details wrong, so check the source.

Named in this story

Gemini 4 Neon, Google DeepMind, Demis Hassabis, Artificial Analysis, Anthropic