Gemini 2.5 Pro is Now #1 on Chatbot Arena with Impressive Jump 🥇

Google DeepMind’s latest AI model, Gemini 2.5 Pro, has reached the #1 position on the Arena leaderboard. The model achieved a notable 40-point score increase over its closest competitors, Grok-3 and GPT-4.5, marking the largest jump ever seen on this leaderboard.

Strong Performance Under Codename “Nebula”

Tested under the codename “nebula,” Gemini 2.5 Pro excelled in all categories evaluated on the Arena leaderboard, earning the top rank across the board. It stood out particularly in Math, Creative Writing, Instruction Following, Longer Query, and Multi-Turn interactions, securing unique #1 spots in these areas. This shows the model’s ability to handle a wide range of tasks, from solving complex math problems to maintaining coherent conversations over multiple turns.

The Arena leaderboard, run by lmarena.ai (formerly lmsys.org), measures how well AI models perform based on human preferences, making Gemini 2.5 Pro’s top ranking a clear sign of its quality and versatility. The 40-point lead over competitors like xAI’s Grok-3 and OpenAI’s GPT-4.5 highlights its strong performance.

A Win for Google DeepMind

Google DeepMind shared that Gemini 2.5 Pro is their “most intelligent model” yet, performing well in math, science, and coding tasks. For example, it scored 18.8% on Humanity’s Last Exam, a tough test of knowledge and reasoning, and showed improvements in coding, such as creating web apps and games.

What is Gemini 2.5 Pro?

Gemini 2.5 Pro, the newest AI model from Google DeepMind, enhances performance, efficiency, and capabilities compared to earlier models. As part of the Gemini 2.5 series, this Pro-tier version delivers a cost-effective balance of power for developers and businesses.

• Multimodal Support: Handles text, images, video, audio, and code, making it versatile across domains.
• Advanced Reasoning: Analyzes information methodically for more accurate, context-aware responses.
• Larger Context Window: Supports 1 million tokens, with plans to expand to 2 million.
• Better Coding: Offers improved code generation and assistance for developers.
• Updated Knowledge: Trained on data up to January 2025.
• Availability: Coming soon to Vertex AI.

For more details on the model, check out our in-depth guide on Gemini 2.5 Pro here!

Looking Ahead

Gemini 2.5 Pro’s success on the Arena leaderboard highlights its strengths in reasoning, coding, and handling complex tasks. It also raises questions about how other AI companies, like OpenAI and xAI, might respond. For now, Gemini 2.5 Pro’s performance sets a new standard, and it will be interesting to see how it shapes the future of AI development.

For more information, check out the full thread on X at lmarena.ai’s post.

Login to continue reading and enjoy expert-curated content.

Free Courses