New Delhi: Google is reportedly preparing to launch a new Gemini artificial intelligence model that could strengthen its position in the increasingly competitive AI coding market. The model, internally codenamed “Skimaki”, could be released as early as September 2, according to a report cited by India Today.

The upcoming model is reportedly called Gemini 3.8 Flash and is expected to focus particularly on coding performance. If the reported results hold up after public testing, the model could put Google in stronger competition with AI systems from Anthropic and OpenAI, which have gained significant ground in software development and coding applications.

However, Google has not publicly confirmed the launch details or the reported performance claims.

Gemini 3.8 Flash could target coding

According to the report, Google has been testing Gemini 3.8 Flash internally and could make it available as early as Wednesday, September 2. The model is said to have been tested by employees over the past month.

Coding has become one of the most commercially important applications of generative AI. Companies are increasingly using AI tools to write, debug and modify software, automate development tasks and operate coding agents.

Anthropic and OpenAI have emerged as particularly strong competitors in this area. Google is therefore putting greater emphasis on coding performance as it works to maintain its position in the rapidly changing AI market.

The reported internal testing has generated particular interest. Some Google engineers reportedly preferred Gemini 3.8 Flash over Anthropic’s Opus model during head-to-head testing on Jetski, Google’s coding platform.

These are internal assessments, however, and should not be treated as independent benchmark results. The model’s actual performance will become clearer only after wider testing and publicly available evaluations.

Flash model designed for speed and lower cost

Gemini 3.8 Flash is expected to be different from Google’s larger flagship models. Flash models are designed to provide faster responses while requiring fewer computing resources than the company’s more powerful Pro models.

The report says Flash models generally use hundreds of billions of parameters, while some of the most advanced models from major AI companies are believed to operate at significantly larger scales. The smaller architecture can also allow Google teams to experiment with different approaches more quickly.

Google has already been accelerating its Flash releases. Gemini 3.6 Flash reportedly arrived in July, followed by Gemini 3.7 Flash only a few weeks later.

This rapid development could allow Google to improve coding capabilities without waiting for a much larger flagship model to be completed.

Google faces pressure in the AI race

The reported Gemini 3.8 Flash launch comes at a crucial time for Google. The company briefly appeared to be among the leaders in the AI race following the release of Gemini 3.0, but newer models from competitors have increased pressure on Google’s AI division.

Google has reportedly experienced delays with Gemini 3.5 Pro after internal candidates did not deliver sufficient improvements over the Flash series, particularly on coding benchmarks.

At the same time, the company’s next major flagship model, Gemini 4, is reportedly performing well during pre-training evaluations but is still undergoing post-training.

This means Google is pursuing several AI development tracks simultaneously, rather than relying on a single model release to regain momentum.

Leadership and research changes add to the pressure

Google’s AI development has also been accompanied by changes among its senior researchers and leadership.

The report says Google co-founder Sergey Brin has urged employees to accelerate Gemini development, particularly after Anthropic released its Claude Mythos model in April.

Several prominent researchers, including Noam Shazeer and Jeff Dean, have also reportedly left the company in recent months.

Meanwhile, Koray Kavukcuoglu has taken over more of the day-to-day responsibilities at Google DeepMind. The report notes that Kavukcuoglu had already been involved in Gemini-related decisions for some time.

Google has also strengthened its research team by hiring Barret Zoph, a former Thinking Machines Lab co-founder and OpenAI post-training lead, as vice-president of research. His work is focused on areas including reinforcement learning and post-training.

Gemini has strong user adoption but faces enterprise competition

Despite the competitive pressure, Google’s Gemini ecosystem has achieved significant user adoption. The Gemini app has reportedly crossed 1 billion users globally.

The bigger challenge is increasingly the enterprise market, where businesses are looking beyond conversational AI towards coding assistants and autonomous AI agents that can complete complex software-development tasks.

This has created a new battleground for Google, OpenAI and Anthropic.

Google’s advantage is its enormous computing infrastructure and broad ecosystem of products and services. Its challenge is translating those advantages into AI systems that developers and businesses prefer over competing tools.

AI coding competition is expanding

The coding competition is no longer limited to Google, OpenAI and Anthropic.

Elon Musk’s xAI has also released newer Grok models aimed at coding-related workloads, while Meta has launched its own AI coding agent, Muse Code, through the team led by Alexandr Wang.

The growing number of competitors suggests that AI coding is becoming one of the most important fronts in the broader generative AI race.

For developers, the competition could eventually mean faster, cheaper and more capable coding assistants. For technology companies, however, winning this segment could translate into significant enterprise adoption and recurring revenue.

What to expect from Google’s new Gemini model

If the reports are accurate, Gemini 3.8 Flash could serve as an important intermediate step for Google while it continues developing its larger flagship models.

The model is not expected to be a new frontier AI system. Instead, its importance could come from delivering strong coding performance while maintaining the speed and lower operating costs associated with Google’s Flash series.

For now, claims that it can outperform Anthropic or OpenAI should be treated cautiously because they are based on reported internal testing rather than independent public benchmarks.

If Google releases the model as reported, wider developer testing and independent evaluations will provide a clearer picture of whether Gemini 3.8 Flash can genuinely challenge the leading AI coding systems.