Google is reportedly preparing to launch Gemini 3.8 Flash, codenamed 'Skimaki,' which aims to reclaim its dominance in the AI race by outperforming competitors in coding tasks.

Loading Video...
  • Google may launch the new Gemini 3.8 Flash model as early as this week.
  • Internally codenamed 'Skimaki,' the model shows superior coding performance in tests.
  • The model is part of the 'Flash' series, optimized for speed and cost-efficiency.
  • Google is intensifying its focus on the enterprise coding market to rival OpenAI and Anthropic.

After losing significant ground to OpenAI and Anthropic in recent months, Google is gearing up for a major comeback. Reports suggest that the tech giant is ready to release a new AI model, internally known as 'Skimaki', which could fundamentally shift the competitive landscape, particularly in the highly lucrative coding domain.

According to insights from the Wall Street Journal, the new model, expected to be branded as Gemini 3.8 Flash, could see a release as early as Wednesday, September 2, 2026. This move comes at a critical juncture for Google, following delays with its flagship Gemini 3.5 Pro and significant leadership transitions within Google DeepMind.

Read Also:Unmanned Revolution: US Fury Drone Successfully Launches Live Air-to-Air Missile

Why This Matters

BozokMedia analysis shows that the battle for AI supremacy has shifted from general conversation to specialized utility. Coding is one of the most significant enterprise use cases for AI. By deploying 'Flash' models—which are smaller, faster, and more cost-effective than trillion-parameter frontier models—Google is targeting the high-volume developer market where latency and cost are decisive factors.

The ability to deliver high-level coding logic within a lightweight, efficient framework will define the next era of AI productivity.

Internal testing on Google’s coding platform, Jetski, has already yielded promising results. Some engineers reportedly preferred the 3.8 Flash model over Anthropic’s highly regarded Opus model. Unlike massive frontier models, the Flash series is trained on a few hundred billion parameters, allowing Google teams to iterate and deploy updates much more rapidly than their competitors.

Historical Background

Google’s journey in the generative AI era has been a rollercoaster. While the release of Gemini 3.0 in November 2025 initially positioned them as leaders, they quickly faced headwinds. The company has dealt with the departure of key researchers like Noam Shazeer and Jeff Dean, and the transition of leadership responsibilities from Demis Hassabis to Koray Kavukcuoglu.

To combat this, Google has aggressively pivoted its resources. The company has hired Barret Zoph, a former OpenAI lead, to spearhead reinforcement learning and post-training efforts, signaling a direct attempt to close the gap in reasoning and coding capabilities.

Did You Know?: Google's Gemini app has already achieved a massive milestone, crossing 1 billion users globally.

Frequently Asked Questions

Question 1: What makes the Gemini 3.8 Flash different from Pro models?
Answer: Flash models are designed to be smaller and faster, offering quicker response times and lower costs compared to the much larger and more resource-intensive Pro models.

Question 2: Who are Google's main rivals in AI coding?
Answer: While OpenAI and Anthropic are the primary rivals, Google also faces competition from Elon Musk's SpaceXAI (Grok) and Meta's Muse Code.