Google Unveils New AI Coding Tools to Challenge Anthropic, OpenAI

Google looks to be taking the fight to Anthropic and OpenAI in the coding space, with a slew of coding and agentic AI announcements at its I/O developer conference.

The search giant’s new model, Gemini 3.5 Flash, is being positioned as the model layer for coding agents. According to Google, it is its strongest agentic and coding model yet, outperforming Gemini 3.1 Pro on key benchmarks and matching or surpassing the latest versions of Claude and GPT. It is doing this at half the cost of comparable models, according to benchmarking data shown by Google at the event.

Cost has become a real pain point for coding and agentic tasks, with major companies burning through their budgets much faster than forecast. Uber recently revealed it had blown through its annual budget for Claude Code in four months. Having a tighter AI model with comparable performance could be exactly how Google gains market share in the enterprise market.

Alongside the new model, Google spent a lot of the conference focused on agent tools. Antigravity 2.0, a platform for managing multiple coding agents, is now a standalone desktop app, and coders can run multiple agents in parallel, allowing them to orchestrate entire workflows rather than simple code completions.

Google also launched Antigravity CLI, the company’s terminal-based answer to Claude Code. With the Antigravity SDK also available to let developers access the same agent frameworks and create custom behaviors, Google is trying to become the full-stack agentic dev platform.

Google has also beefed up its vibe coding studio, with developers now able to code Android apps natively through prompts. These can be previewed in an Android emulator and published to Google Play’s Internal Test Track, essentially creating an automated step-by-step path from idea to storefront.

The pivot from users to enterprise customers

Google has been trying to pivot toward enterprise customers for a few months now. In March, it showed that Gemini beat Claude and GPT on certain coding tasks, although the industry still seems broadly infatuated with Claude. It also unveiled the Gemini Enterprise Agent Platform last month in Las Vegas.

It is growing its enterprise customer base at a fast clip. “Two years ago, we were processing 9.7 trillion tokens a month across our surfaces — a huge number,” said Pichai. “Last year at I/O, that grew to roughly 480 trillion tokens. Fast forward to today, that number jumped 7x to over 3.2 quadrillion per month.”

The shift in focus to enterprise is partly due to the surge in revenue Anthropic has seen since the start of 2026, a decent portion of which has come from its coding and agentic tools. While ChatGPT and Gemini both have 900 million active users, Anthropic has reportedly matched OpenAI’s revenue run rate through its growth in enterprise. 

While Google does not break out its AI revenue, there have been reports of lower monetization per user than Anthropic and GPT for coding and agentic tasks, and it is in third place for market share, according to Menlo Ventures.

Other AI model vendors are also looking to get a piece of the lucrative coding market. xAI launched Grok Build this week, a few weeks after CEO Elon Musk internally pushed for more enterprise tools to compete with Claude and GPT.

Even though Google may be behind Anthropic and OpenAI on the coding and agentic front, it is more than making up for this with increased ad revenue, Google Cloud customers, and Workspace. The company improved its year-on-year revenue by 22 percent in the first quarter of this year, with AI contributing a lot to this performance. 

Also read: Google and Blackstone are creating a $5 billion TPU cloud venture to expand enterprise access to AI compute.

The post Google Unveils New AI Coding Tools to Challenge Anthropic, OpenAI appeared first on eWEEK.