Skip to main content
Google's Gemini 4 Neon AI model with one-million-token context, represented by a glowing brain and neural network.

Editorial illustration for Google's Gemini 4 Neon AI Model Has One-Million-Token Context

Google's Gemini 4 Neon Hits 1M Token Context Window

Google's Gemini 4 Neon AI Model Has One-Million-Token Context

• 4 min read

Google DeepMind shipped its fourth AI model in seven weeks on Wednesday, a pace that would have seemed implausible even a year ago. The new release, Gemini 4 Neon, slots into a lineup that already includes Carbon and Hydrogen, and it's built to handle a one-million-token context window across text, images, audio and video. That's the same ceiling Carbon offers, but Neon gets there faster, running at roughly 110 tokens per second, nearly double Carbon's speed.

On the Artificial Analysis Intelligence Index, Neon's top reasoning setting, labeled "xhigh," pulls in 61 points, just ahead of Carbon's 60. That's enough to land third place overall, behind OpenAI's GPT-6.2 Astra and Anthropic's Claude Fable 5.2, though Google leaves both rivals out of its own comparison charts. Neon also claims wins on coding benchmarks against Anthropic's competing model, at a fraction of the cost, and holds up well in image analysis and safety testing.

The pricing structure behind that cost advantage is where things get interesting, and it has less to do with how efficiently Neon uses tokens than with where Google set the sticker price in the first place.

Google DeepMind has released Gemini 4 Neon, its fourth model in seven weeks. Neon is nearly twice as fast as its sibling Carbon, beats it on most benchmarks and costs significantly less.

Why this matters

Four models in seven weeks tells you Google is optimizing for release velocity, not just capability. For developers, the pitch is straightforward: a million-token context window and coding performance that beats Anthropic's model at a fraction of the cost, running at 110 tokens per second. That's a real reason to swap API calls today, not just a benchmark headline.

But third place on Artificial Analysis, behind OpenAI and Anthropic, is worth sitting with. Google is winning on price and speed, not on raw intelligence, at least by that measure. If you're building products where cost per token and context length matter more than squeezing out the last few points on a reasoning benchmark, Neon is the obvious pick right now.

If you need the absolute ceiling on reasoning, it isn't that yet. The bigger signal is the cadence itself: Carbon to Hydrogen to Neon in seven weeks means whatever you integrate this quarter will likely be outdated by the next one. Plan your infrastructure for churn, not permanence.

Common Questions Answered

How does Gemini 4 Neon's processing speed compare to Google's Carbon model?

Gemini 4 Neon runs at approximately 110 tokens per second, which is nearly double the speed of Carbon. Despite matching Carbon's one-million-token context window ceiling, Neon achieves this faster performance while also costing significantly less.

What multimodal capabilities does Gemini 4 Neon support with its one-million-token context window?

Gemini 4 Neon is built to handle a one-million-token context window across text, images, audio, and video. This multimodal support allows developers to process diverse content types within the same extended context window that Carbon offers.

What is Google DeepMind's release strategy demonstrated by Gemini 4 Neon?

Google DeepMind released Gemini 4 Neon as its fourth AI model in just seven weeks, indicating a focus on optimizing for release velocity rather than solely on capability improvements. This rapid release pace would have seemed implausible just a year ago and reflects Google's strategy to maintain competitive momentum in the AI market.

How does Gemini 4 Neon's cost and performance compare to Anthropic's models?

Gemini 4 Neon beats Anthropic's model on coding performance benchmarks while running at 110 tokens per second and costing significantly less. This combination of superior performance, speed, and lower cost provides developers with a compelling reason to switch API calls from competitors.

LIVE17:16Infosys Invests USD 10.5 Million in NTT DOCOMO Metaverse Venture