Home · Technology · Nov 21 archive

Google Launches Gemini 4 Neon for AI Pro Users

Confirmed

Technology Desk

In Short: Neon is accessible to both Pro and Ultra subscribers, lowering the barrier to entry compared to previous models like Carbon, which were restricted to Ultra users.

Google; Google Gemini logos
Logos: Google (Public domain); Google Gemini (Public domain)

Google today launched Gemini 4 Neon, the newest addition to its Gemini series, available for AI Pro and Ultra subscribers via the Gemini API and Google Antigravity platform.

Neon is positioned as a high-performance model designed for enterprise, scientific, and software development tasks, ranking third overall on the Artificial Analysis Intelligence Index with a score of 61 at its xhigh thinking level.

In Google Antigravity, Neon demonstrated its capabilities by creating a first-person WebGL2 freeride ski simulator from scratch, incorporating real Mont Blanc and Chamonix terrain data.

Neon excels in multi-step instruction following and has shown strong performance in benchmarks across finance, accounting, spreadsheet modeling, and complex constraint following.

SVP and Chief AI Architect at Google, Google Today, announced the release of Neon, emphasizing its ability to complete end-to-end tasks quickly and efficiently.

Neon's pricing is set at $3 per million input tokens and $15 per million output tokens, with a 95% discount for cached input tokens.

Neon is accessible to both Pro and Ultra subscribers, lowering the barrier to entry compared to previous models like Carbon, which were restricted to Ultra users.

Google has also introduced a new AI Ultra Lite tier, priced between the $20 Pro plan and the $250 Ultra plan, offering a middle ground for heavy users.

Neon's release follows the launch of Argon and Carbon earlier this year, continuing Google's strategy of releasing powerful AI models.

Neon's capabilities extend to enterprise work, where it can handle spreadsheets, PDFs, charts, audio, and video, making it a versatile tool for complex workflows.

What's confirmed

What's still developing

Sources