Google Says Gemini 2.5 Pro Outperforms DeepSeek R1 and Grok 3 Beta in Coding Benchmarks

Google has unveiled a refreshed preview of its flagship Gemini 2.5 Pro model, initially introduced in March and enhanced in May. This iteration, described as the company's "most intelligent" AI to date, is currently in preview with plans for general availability within weeks.

Businesses can now experiment with developing new applications or upgrading existing implementations using the updated "I/O edition" of Gemini 2.5 Pro. According to Google's official announcement, this version delivers more imaginative responses and demonstrates superior capabilities in programming and logical reasoning compared to previous iterations.

Our latest Gemini 2.5 Pro update is now in preview.

It's better at coding, reasoning, science + math, shows improved performance across key benchmarks (AIDER Polyglot, GPQA, HLE to name a few), and leads @lmarena_ai with a 24pt Elo score jump since the previous version.

We also… pic.twitter.com/SVjdQ2k1tJ

— Sundar Pichai (@sundarpichai) June 5, 2025

At its May developer conference, Google revealed enhancements to Gemini 2.5 Pro that surpassed its earlier silent release. Demis Hassabis, CEO of Google DeepMind, highlighted the I/O edition as the company's most advanced coding model to date.

This newest preview, designated Gemini 2.5 Pro Preview 06-05 Thinking, advances beyond the I/O edition's capabilities. The forthcoming public release promises enterprise-grade performance and scalability.

The original I/O edition (gemini-2.5-pro-preview-05-06) became accessible to developers and corporations in May via Google AI Studio and Vertex AI. The enhanced Gemini 2.5 Pro Preview 06-05 Thinking is available through these same channels.

Performance metrics

This upgraded Gemini 2.5 Pro demonstrates measurable improvements over its predecessor.

Google reported a 24-point advancement in LMArena and a 35-point gain in WebDevArena, where it now leads competitor rankings. Comparative testing revealed superior performance against models including OpenAI's o3, o3-mini, and o4-mini, Anthropic's Claude 4 Opus, xAI's Grok 3 Beta, and DeepSeek R1.

"We've also addressed feedback from our previous 2.5 Pro releases, improving its style and structure — it can be more creative with better-formatted responses," Google stated in its announcement.

What enterprises can expect

While Google's rapid Gemini 2.5 Pro enhancements may seem complex, the company positions them as direct responses to user input. The new version carries a pricing structure of $1.25 per million input tokens (without caching) and $10 per million output tokens.

When Gemini 2.5 Pro debuted in March, industry observers recognized it as an underutilized advanced model. Google has since embedded the technology across numerous applications, including the "Deep Think" feature that evaluates multiple hypotheses before generating responses.

Gemini 2.5 Pro's release and subsequent upgrades have reinforced Google's position in the competitive large language model landscape, reclaiming attention from rival reasoning models by DeepSeek and OpenAI.

Within hours of the announcement, developers began testing the updated Gemini 2.5 Pro. Early impressions confirm Google's claims of accelerated performance, though comprehensive evaluation of its enhanced capabilities remains ongoing.

First hour with "Gemini 2.5 Pro Preview 06-05"

Positives:

– It's faster
– It produces more output
– It has a better macro play (multi file edits, better overview)
– Output structure is better (readable)
– It's more concise and LESS APOLOGETIC!!

Before: "You are absolutely…

— Patrick Bade (@nishffx) June 5, 2025

you guys cooked, really enjoying the app builder.

made a game and tested it out, it was using imagen to build assets on the fly ? and it's up, hosted, easy to share. Really the best no-experience no-code builder yet.

keep building out the vibe app marketplace, this could…

— bone (@boneGPT) June 5, 2025

Gemini 2.5 Pro Preview is pretty good.. used it yesterday for deep research and the results are better than some of the big names..

— Janak (@janaks09) June 5, 2025