All posts

Gemini's market share dropped to 1.9% as Claude surges

Huma ShaziaAugust 14, 2026 at 12:02 AM4 min read
Gemini's market share dropped to 1.9% as Claude surges

Google's Gemini dropped from 12% to 1.9% market share between early 2026 and July, according to AI text recognition platform Pangram. The collapse, confirmed by separate OpenRouter data, coincides with Anthropic's Claude climbing from 4.3% to 14.9% in the same period. OpenAI's ChatGPT has held above 50% every month since Pangram began tracking.

Three independent data sources now point in the same direction: users are choosing Gemini less often, even as Google claims one billion monthly active users for its Gemini App.

1.9%
Gemini's July 2026 market share, down from 12% earlier in the year (Pangram data)
Advertisements

What the data actually measures

Google Gemini Loses Market Share to ChatGPT & Claude

Pangram tracks model usage by analyzing submitted texts. Its data skews toward writing tasks, where Gemini has historically trailed both Claude and ChatGPT. OpenRouter, which routes API calls to various models, skews toward open-weight models and developer workflows. Similarweb measures website traffic and ignores mobile apps entirely.

Chart comparing Pangram and OpenRouter market share data showing Gemini's decline
Image (Source: The Decoder)

Each source has blind spots. But when three different methodologies tell the same story, the convergence matters more than any single metric.

Similarweb data shows Google's website market share dipped last month from 27% to 26.8%. That figure was just 9.4% a year ago, so the year-over-year growth remains strong even as the monthly trend flattens.

Claude's gains in technical fields

Anthropic's growth came primarily in technical and scientific fields, per Pangram. That tracks with Claude's reputation for extended context handling and coding tasks. Product teams building AI-powered features have increasingly turned to Claude for agentic workflows, where instruction-following and reasoning matter more than raw speed.

The shift creates a practical problem for teams that standardized on Gemini early. Switching models mid-project means rewriting prompts, re-testing edge cases, and often renegotiating API contracts. For startups burning runway, the switching cost competes with the performance gain.

Also Read
7 Claude Code alternatives for different workflows

Relevant for teams evaluating Claude's ecosystem against alternatives

Google's billion-user claim

Google says one billion people use the Gemini App monthly. Monthly active users is a weak signal. Opening an app once because it shipped pre-installed on Android counts the same as daily power use. The Pangram and OpenRouter data measure something different: which model people choose when they have a choice.

That gap between install-base metrics and usage-share metrics matters for product teams deciding which API to integrate. A large user base does not guarantee the model performs well enough to retain users who actively shop alternatives.

Leadership changes at DeepMind

The market share decline may help explain Google DeepMind's recent leadership shakeup. Several senior figures departed, including Demis Hassabis, who co-founded DeepMind in 2010 and led the merger with Google Brain in 2023.

Whether the departures drove the decline or the decline drove the departures, the timing is not coincidental. Google's AI strategy now faces pressure on two fronts: retaining research talent and regaining competitive share in a market where OpenAI and Anthropic have established durable leads.

ℹ️

Logicity's Take

For AI product teams, this data changes the vendor calculus. Gemini's integration with Google Cloud and Workspace remains a draw, but the performance gap in writing and coding tasks is hard to ignore. Teams evaluating models should benchmark on their actual use case, not aggregate market share. If your workflow leans technical, Claude's surge suggests you test it seriously. OpenAI still dominates volume, but Anthropic is capturing the margin.

The next quarter will show whether Google can stabilize. New model releases, pricing changes, or enterprise deals could shift the trajectory. For now, the data says users are voting with their API calls, and Gemini is losing.

ℹ️

Need Help Implementing This?

If you're evaluating AI models for production workloads or need help migrating between providers, reach out to the Logicity team. We help engineering teams make sense of the vendor landscape and ship faster.

Source: The Decoder / Matthias Bastian

H

Huma Shazia

Senior AI & Tech Writer

Produced with AI assistance and reviewed by the Logicity editorial team. Learn more in our Editorial Policy.