Gemini 4 Neon: Google DeepMind’s Frontier Coding Agent
Koray Kavukcuoglu, Google DeepMind SVP, details how the new model Gemini 4 Neon for long-running agent workflows tops coding benchmarks at lower task costs

TL;DR
- Google DeepMind released its fourth new AI model, Gemini 4 Neon, in seven weeks.
- Gemini 4 Neon ranks third overall on the Artificial Analysis Intelligence Index with a score of 61 at its xhigh thinking level.
- It supports a 1 million token context window and costs $3 per million input tokens and $15 per million output tokens.
- The model excels in software engineering, scientific reasoning, analytical workflows, coding, and multi-step instruction following.
- Gemini 4 Neon ranks second on the Vals Index for potential economic impact and second on steerability.
- Customers report Gemini 4 Neon offers a balance of quality, cost, and speed, matching or exceeding previous models at a lower price and higher speed.
- Google DeepMind is making a comeback at the AI frontier, with key partners including Apple, Samsung, and Accenture.