[KongchangAI]
ModelGemini 4 Neon / Neon

Gemini 4 Neon

A model in Google's Gemini 4 series whose specific positioning remains unclear; it may be an upgraded version or part of a new series.

Core Facts

Timeline (last 90 days)

Nov 19

如果 Gemini 4 Neon 具备与 Opus 5.5 竞争的实力,开发者的模型选型将更依赖具体任务表现,而非品牌惯性。

Unverified50%
Nov 19

「Gemini 4 Neon」的命名带有推测成分,尚无官方确认。

Unverified50%
Nov 19

Hacker News 上出现了一条标题为「Gemini 4 Neon looks like an Opus 5.5 competitor」的讨论帖,获得 18 分和 1 条评论。

Unverified50%
Nov 19

目前关于 Gemini 4 Neon 与 Opus 5.5 对比的所有判断均建立在一条简短帖子的标题之上,缺乏可验证的基准数据或产品细节。

Unverified50%
Nov 19

Before the official documentation is completed, using Gemini 4 Neon for critical business systems still carries risk.

Unverified50%
Nov 19

The official blog post for Gemini 4 Neon is categorized under the "Innovation & AI / Models & Research" section.

Unverified50%
Nov 18

Gemini 4 Neon offers four reasoning levels: low, medium, high, and xhigh.

Unverified50%
Nov 18

The official documentation explicitly lists coding, agent workflows, and multi-step tasks as key focus areas for Gemini 4 Neon.

Unverified50%
Nov 18

The model ID for Gemini 4 Neon in AI Gateway is google/gemini-4-neon.

Unverified60%
Nov 18

Gemini 4 Neon supports tool calling, structured outputs, and grounding capabilities based on Google Search and Google Maps.

Unverified50%

17 more timeline events

All Facts (20)

Verified

Google published a new member of the Gemini series, Gemini 4 Argon, on its official blog.

90%
Partially Verified

The identifiers Gemini 4 Neon and Gemini 4 Oxygen appear in Gemini Web backend logs, alongside Gemini 4 Argon and Carbon.

75%
Unverified

The model ID for Gemini 4 Neon in AI Gateway is google/gemini-4-neon.

60%
Unverified

目前关于 Gemini 4 Neon 与 Opus 5.5 对比的所有判断均建立在一条简短帖子的标题之上,缺乏可验证的基准数据或产品细节。

50%
Unverified

「Gemini 4 Neon」的命名带有推测成分,尚无官方确认。

50%
Unverified

Before the official documentation is completed, using Gemini 4 Neon for critical business systems still carries risk.

50%
Unverified

The official blog post for Gemini 4 Neon is categorized under the "Innovation & AI / Models & Research" section.

50%
Unverified

Gemini 4 Neon supports tool calling, structured outputs, and grounding capabilities based on Google Search and Google Maps.

50%
Unverified

Gemini 4 Neon offers four reasoning levels: low, medium, high, and xhigh.

50%
Unverified

The official documentation explicitly lists coding, agent workflows, and multi-step tasks as key focus areas for Gemini 4 Neon.

50%
Unverified

Gemini 4 Neon supports streaming calls via the AI SDK's streamText.

50%
Unverified

On SWE-sweep, Gemini 4 Neon scores 12.8% under the mini-SWE harness and 39.0% in Google Antigravity.

50%
Unverified

Gemini 4 Neon sits on the Pareto frontier on Vals Index and Agent Arena, with cost per task about 50% to 80% lower than top competitors.

50%
Unverified

Gemini 4 Neon scored 61 on the Artificial Analysis Intelligence Index at the xhigh thinking level, ranking third overall.

50%
Unverified

Gemini 4 Neon ranked second on the Agent Arena Index with a score of 1336, an evaluation covering more than 4,200 real agent sessions.

50%
Unverified

Gemini 4 Neon ranked first on both Legal Research Bench and Tax Agent Bench.

50%
Unverified

In Surge AI's evaluation, Gemini 4 Neon ranked first in multiple categories including finance, accounting, spreadsheet modeling, and complex constraint following.

50%
Unverified

Gemini 4 Neon scored 56.3% on Zapier's AutomationBench and $16,843 on Andon Labs' Vending-Bench 2 (a long-horizon business simulation), taking first place overall on both.

50%
Unverified

Gemini 4 Neon scores 70.5% on Terminal-Bench Science, 84.0% on RiemannBench, and 71.3% on HLE-Diamond, leading in all.

50%
Unverified

Gemini 4 Neon ranked second on the Vals Index with a score of 70.9%.

50%

Source Articles