Gemini API: Gemini 4 Neon Reaches General Availability With Four Thinking Levels and a 1M-Token Context Window

Gemini CLIView original changelog

Google made Gemini 4 Neon (gemini-4-neon) generally available in the Gemini API, positioning it as a model with frontier-level intelligence and fast end-to-end task completion for software engineering, knowledge work and science. It supports four thinking levels, including xhigh, and a unified 1M-token context window with long decode up to 1M tokens. Paid-tier pricing is $3.00 per 1M input tokens and $15.00 per 1M output tokens.

Key Takeaways

  • Gemini 4 Neon is now generally available in the Gemini API as gemini-4-neon, so it can be used in production rather than as a preview.
  • Google positions the model for complex software engineering, knowledge work and science workflows, with an emphasis on fast end-to-end task completion.
  • Four thinking levels (low, medium, high and xhigh) let developers trade reasoning depth against latency and cost, with high as the default.
  • A unified 1M-token context window covers inputs large enough for whole codebases or document collections.
  • Long decode up to 1M tokens allows very long continuous outputs in a single response.
  • Paid-tier pricing of $3.00 per 1M input and $15.00 per 1M output tokens makes the cost of high thinking levels worth budgeting for.

Gemini 4 Neon Is Generally Available

Google announced general availability of Gemini 4 Neon, available in the Gemini API under the model ID gemini-4-neon. Google describes the model as bringing frontier-level intelligence with fast end-to-end task completion, aimed at complex software engineering, knowledge work and science workflows. For developers who build coding agents and other long-running automation, the emphasis on finishing tasks quickly from start to finish is the headline.

Thinking Levels

Gemini 4 Neon supports four thinking levels: low, medium, high and xhigh. The default is high. Developers can pick low or medium when latency and cost matter more, and move to xhigh for the hardest problems where extra reasoning is worth the wait. The same level names are used on other recent Gemini 4 models, so existing configuration should carry over with little change.

Context Window and Output Length

The model uses a unified 1M-token context window and supports long decode of up to 1M tokens. A single window for input and a very large ceiling for generated output make it practical to feed in large codebases or document sets and to ask for long, continuous results such as big refactors or extended reports.

Pricing

On the paid tier, Gemini 4 Neon costs $3.00 per 1M input tokens and $15.00 per 1M output tokens. Google points developers to its "What's new in Gemini 4 Neon" guide for getting started.

What It Means for Developers

Tools that call the Gemini API, including Gemini-powered coding assistants, can now target gemini-4-neon as a generally available model rather than a preview. Teams comparing models should weigh the mid-range per-token pricing against the choice of thinking level, since the level chosen largely determines how much reasoning, and therefore how many output tokens, a request consumes.