Google DeepMind 发布 Gemini 4 Neon:100 万上下文,登顶 AA 编程智能体指数
Google DeepMind 刚刚发布了 Gemini 4 Neon。Gemini 4 Neon 是一款推理模型,定位介于 Gemini 4 Carbon 和 Gemini 4 Hydrogen 之间。它面向编程智能体、知识工作和科研工作流。面向开发者的卖点是:以更低的单任务成本,实现接近前沿水平的得分。定价为每 100 万输入 Token 3 美元,每 100 万输出 Token 15 美元。
Neon 是 Gemini 4 家族中的中间档模型。Carbon 负责难度最高的长周期项目,Hydrogen 负责快速的日常执行,Neon 则在两者之间平衡深度与效率。
根据模型卡,Neon 基于 Gemini 4 Carbon 适配而来。Google 未披露参数量或架构。输入支持文本、图像、音频和视频。知识截止时间为 2026 年 3 月。
Neon 提供多个思考等级。High 为默认等级,xhigh 则保留给难度最高的工程、分析和科研工作负载。
针对长时间运行的智能体任务,Neon 支持上下文压缩,并减少冗余的文件读取。在 Google Antigravity 中,Neon 可以充当编排者,指挥并行的 Gemini 4 Hydrogen 子智能体处理更小的子任务。
Google 将 Neon 与 Claude Opus 5.5 和 GPT-6.2 Sol 进行对比。以下所有得分均由 Google 报告,并非独立复现的结果。
Neon 并非在每一项上都获胜。Opus 5.5 在 Terminal-bench 4.1(65.8% 对 65.2%)和 FrontierCode 2.0(47.3% 对 46.2%)上领先。GPT-6.2 Sol 在 Chartography(75.1% 对 73.5%)上领先。
在 Vals Index 上,Neon 得分为 70.9%,排名第 2。在 Agent Arena 指数上它同样排名第 2,得分为 1336。
Neon 的标价为每 100 万 Token 输入 $3、缓存输入 $0.15、输出 $15。谷歌表示,Neon 在 high thinking 模式下每项任务成本比 Carbon 低 32%。谷歌还宣称,在 Vals Index 和 Agent Arena 上,其每项任务成本比竞争对手低 50% 到 80%。在 xhigh 模式下,谷歌给出的数据是比竞争对手在其最高设置下的成本低 36% 以上。
这些都是按任务计算的成本说法,而非每 Token 的价格。GPT-6.2 Sol 标价为 $2/$10。Claude Opus 5.5 标价为 $4/$20。Gemini 4 Carbon 在 11 月 30 日前为 $2/$10,之后为 $4/$20。
Neon 运行在 Gemini 4 的安全与防护技术栈之上。谷歌报告称,其智能体安全套件上的失准行为率为 0.09%。它通过了 99.2% 的智能体诚实性评估。在最高思考等级下,它在 Gray Swan 的提示注入基准测试中攻击成功率为 0.0%。在编程作弊测试中,其 2.45% 的捷径率为所有 Gemini 4 模型中最低。
*分数来自谷歌发布会对比表,为自报数据。Opus 5.5 和 GPT-6.2 Sol 的 AA 分数与 Artificial Analysis 一致。
请查看发布博客和模型卡。你可以通过我们的 GPULeaderboard.com Chrome 扩展在此查看部署选项和规格。
需要与我们合作推广你的 GitHub 仓库、Hugging Face 页面、产品发布或网络研讨会等?请联系我们。
文章《Google DeepMind 发布 Gemini 4 Neon:100 万上下文,在 AA Coding Agent Index 上排名第一》首发于 MarkTechPost。
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(Google DeepMind Releases Gemini 4 Neon: 1M Context, #1 on AA Coding Agent Index)
Google DeepMind just released Gemini 4 Neon. Gemini 4 Neon is a reasoning model that sits between Gemini 4 Carbon and Gemini 4 Hydrogen. It targets coding agents, knowledge work and science workflows. The developer pitch is near-frontier scores at a lower cost per task. Pricing is $3 per 1M input tokens and $15 per 1M output tokens.
TL;DR
- Size: 1M-token unified context, 64K default output, up to 1M via background long decode.
- Runs on: Google-hosted only. Closed weights, so no local GPU, quantized or edge builds.
- Performance: 61 on the AA Intelligence Index, ahead of Claude Opus 5.5 (59) and GPT-6.2 Sol (57).
- Best: 95.2% on GraphWalks (256K to 1M context), vs 80.1% for GPT-6.2 Sol.
- Worst: 46.2% on FrontierCode 2.0, behind Claude Opus 5.5 at 47.3%.
- Bottom line: Best: leads long-context and agentic coding at $3/$15. Worst: closed, self-reported, pricier per token than GPT-6.2 Sol.
What is Gemini 4 Neon?
Neon is the middle model in the Gemini 4 family. Carbon handles the hardest long-horizon projects. Hydrogen handles fast everyday execution. Neon balances depth and efficiency between the two.
As per the model card, Neon is adapted from Gemini 4 Carbon. Google does not disclose parameter count or architecture. Inputs cover text, images, audio and video. The knowledge cutoff is March 2026.
How does Neon work inside agent workflows?
Neon exposes multiple thinking levels. High is the default. xhigh is reserved for the hardest engineering, analytical and science workloads.
For long agent runs, Neon supports context compaction and makes fewer redundant file reads. In Google Antigravity, Neon can act as an orchestrator. It directs parallel Gemini 4 Hydrogen subagents on smaller subtasks.
How does Gemini 4 Neon perform on benchmarks?
Google compares Neon against Claude Opus 5.5 and GPT-6.2 Sol. All scores below are reported by Google, not independently reproduced.
- Coding: #1 on the AA Coding Agent Index at 58.6. SWE-sweep with provider harness hits 39.0%, vs 10.1% for Opus 5.5.
- Knowledge work: 56.3% on Zapier AutomationBench, vs 42.5% for Opus 5.5. 75.8% on Agents’ Last Exam v1.1.
- Science and reasoning: 84.0% on RiemannBench, 70.5% on Terminal-Bench-Science and 71.3% on HLE-Diamond.
- Long context: 95.2% on GraphWalks, vs 66.8% for Opus 5.5.
- Cyber: 90.5% on SEC-bench Pro, 70.7% on CWE-Bench v2 and 51.3% on WebExploitBench v1.2.
Neon does not win every row. Opus 5.5 leads Terminal-bench 4.1 (65.8% vs 65.2%) and FrontierCode 2.0 (47.3% vs 46.2%). GPT-6.2 Sol leads Chartography (75.1% vs 73.5%).
On the Vals Index, Neon scores 70.9% and ranks #2. On the Agent Arena index it also ranks #2, with a score of 1336.
What does Gemini 4 Neon cost?
Neon lists at $3 input, $0.15 cached input and $15 output per 1M tokens. Google says Neon at high thinking costs 32% less per task than Carbon. It also claims 50% to 80% lower cost per task vs rivals on Vals Index and Agent Arena. At xhigh, Google cites 36%+ lower cost than rivals at their highest settings.
Those are per-task claims, not per-token prices. GPT-6.2 Sol lists at $2/$10. Claude Opus 5.5 lists at $4/$20. Gemini 4 Carbon is $2/$10 until November 30, then $4/$20.
Where can developers use Gemini 4 Neon?
- Gemini API paid tier, via Google AI Studio and Android Studio
- Gemini for business, via API and the model dropdown
- Google Antigravity, for paid Google AI Pro and Ultra subscribers
- The Gemini app, for Pro and Ultra subscribers
What do the safety numbers show?
Neon runs on the Gemini 4 safety and security stack. Google reports a 0.09% misaligned action rate on its agentic safety suite. It passes 99.2% of the agent honesty evaluation. It shows a 0.0% attack success rate on Gray Swan’s prompt injection benchmark at the highest thinking level. Its 2.45% shortcut rate on a coding cheating test is the lowest of any Gemini 4 model.
Gemini 4 Neon vs closest competitors
| Feature | Gemini 4 Neon | Gemini 4 Carbon | Claude Opus 5.5 | GPT-6.2 Sol |
|---|---|---|---|---|
| Developer | Google DeepMind | Google DeepMind | Anthropic | OpenAI |
| Parameters | Not disclosed | Not disclosed | Not disclosed | Not disclosed |
| Weights | Closed | Closed | Closed | Closed |
| Context window | 1M | Not disclosed in launch post | 1M | 1M |
| Max output | 64K default, 1M background | Not disclosed | 128K | Not disclosed |
| Input / output per 1M | $3 / $15 | $2 / $10 to Nov 30, then $4 / $20 | $4 / $20 | $2 / $10 |
| AA Intelligence Index | 61 | Not disclosed | 59 | 57 |
| Vals Index v2.2* | 70.9% | Not disclosed | 70.1% | 66.1% |
| Terminal-bench 4.1* | 65.2% | Not disclosed | 65.8% | 61.5% |
| SWE-sweep (provider harness)* | 39.0% | Not disclosed | 10.1% | 8.3% |
| GraphWalks 256K to 1M* | 95.2% | Not disclosed | 66.8% | 80.1% |
*Scores from Google’s launch comparison table, self-reported. AA scores for Opus 5.5 and GPT-6.2 Sol match Artificial Analysis.
Key Takeaways
- Neon sits between Carbon and Hydrogen, priced at $3/$15 per 1M tokens.
- Scores 61 on AA Intelligence Index, #1 on AA Coding Agent Index.
- Biggest lead: 95.2% on GraphWalks long-context recall.
- Trails Opus 5.5 on Terminal-bench 4.1 and FrontierCode 2.0.
- Closed and API-only, with Google-reported benchmarks.
Check out the launch post and the model card. You can check the deployment options and specs here through our GPULeaderboard.com chrome extension.
Also, feel free to follow us on Twitter and don’t forget to join our 150k+ML SubReddit and Subscribe to our Newsletter. Wait! are you on telegram? now you can join us on telegram as well.
Need to partner with us for promoting your GitHub Repo OR Hugging Face Page OR Product Release OR Webinar etc.? Connect with us
The post Google DeepMind Releases Gemini 4 Neon: 1M Context, #1 on AA Coding Agent Index appeared first on MarkTechPost.
正文保留英文原文,标题与摘要由机器翻译。
原文发布于 MarkTechPost(Google DeepMind Releases Gemini 4 Neon: 1M Context, #1 on AA Coding Agent Index),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。