The artificial intelligence scene is experiencing a profound revolution in June 2025. The emergence of high-performing models is redefining the standards of technological innovation. Companies and researchers are showcasing themselves in fierce competition, driving the quality of solutions offered on the market. The establishment of an accurate ranking allows for the identification of leaders and those emerging on the horizon. This ranking adheres to rigorous criteria, based on anonymous evaluations to ensure impartiality. The stakes are high, as behind each high-performing model lies a potential impact on various industrial sectors.
Recent advances deserve particular attention. Users must necessarily inform themselves about those shaping the digital future. Detailed rankings thus provide an essential overview of current trends in AI.
Overall Ranking of AI Models
The overall ranking of LMArena for June 2025 places Gemini-2.5-pro-preview at the top. This model proves exceptional in programming, mathematics, and creative writing.
OpenAI follows with its model o3, which, despite solid performance, falters on the longest queries. Claude-Opus-4, a model from Anthropic, reaches the sixth position, demonstrating a significant advance compared to its previous rankings.
Text Category
Performance in writing and rephrasing is imperative. In this category, Gemini and ChatGPT compete for the top spot, their results nearly overlapping.
The other models show competitiveness, but none reach the mastery level of this duo, reflecting a renewed battle between two giants of AI.
Web Development Category
In web development, Gemini maintains its dominant position despite notable changes in the hierarchy. DeepSeek and several Claude models emerge in the standings, but Gemini’s supremacy persists.
This sector demands heightened skills to correct, write or explain code, which remains the strong point of the leading models.
Image Analysis Category
The ranking of models specialized in image analysis reveals an absence of significant competition for Google and OpenAI. Both companies occupy the entirety of the top 10, illustrating their undeniable expertise in this field.
This dominance highlights the relevant blind spots in previous LMArena evaluations, where model diversity was not sufficiently taken into account.
Online Research Category
Online research benefits from particular attention in the ranking. Google, with the Gemini-2.5-grounding model, rises to the top thanks to its ability to simulate searches for relevant information. Perplexity, with its Sonar engine, also emerges, strengthening the competition.
OpenAI maintains a less enviable position, its presence being hindered by its API versions of GPT-4o, confined to lower ranks.
Code Assistants Category
The diversity of players in the code assistants field is striking. Models such as DeepSeek, Claude, and Mistral vie for supremacy. No single company seems to dominate this segment, leaving room for a varied competitive landscape.
This plurality of offerings paves the way for innovative solutions, pushing models to effectively respond to diverse development assistance requests.
Image Generation Category
OpenAI, with GPT-Image-1, achieves a significant breakthrough in the field of image generation. This model significantly surpasses those from Google, which, although well-established in the ecosystem, see their position challenged.
Competitors such as Ideogram, Recraft, and Stable Diffusion also stand out for their ability to create quality visuals, enriching the creative landscape of AI.
New Ranking Criteria and LMArena
LMArena, an open-source platform, introduces significant changes to its evaluation of AI models. Users participate in anonymous comparisons, assigning an Elo score to each model based on duels.
This innovative system helps facilitate a fairer and more balanced assessment of performances, relying on a dynamic ranking that takes into account the specifics of each model.
Frequently Asked Questions About the 20 Most Performant AI Models in June 2025
What criteria were used to establish the ranking of AI models in June 2025?
The ranking is based on several criteria, including the ability to understand and generate text, web development aptitude, image analysis, online research, code assistance, and image generation.
Which AI model was crowned the best in June 2025?
The model that achieved the first place in the overall ranking is Gemini-2.5-pro-preview, which excels in several areas such as coding, mathematics, and creative writing.
How are the performances of AI models measured in LMArena?
Performances are assessed through anonymous evaluations made by users, who choose which model provided the best response for each prompt, thereby influencing the Elo score of each model.
What are the main domains in which AI models are ranked in LMArena?
AI models are ranked in six main domains: text, web development, image analysis, online research, code assistants, and image generation.
Why did LMArena change its name and add several rankings?
LMArena expanded its name to better reflect the diversity of AIs and to segment the overall ranking into six distinct tables, allowing for a better appreciation of model performance in specific domains.
Which models have shown remarkable performance in web development?
In the web development category, Gemini continues to dominate, while other models like DeepSeek and several variants of Claude also appear in the lead pack.
Which model stood out in image analysis in June 2025?
Google and OpenAI dominate without competition in the image analysis category, with their models occupying the entire top 10.
How do the changes made to LMArena influence the competition between AI models?
The changes, notably the addition of specific rankings, allow for a finer assessment of model capabilities, leading to a more balanced competition and highlighting the strengths and weaknesses of each AI.
Have Anthropic’s AI models improved compared to past rankings?
Yes, Anthropic has seen significant improvement with the Claude-Opus-4 model, rising in the rankings after previously stagnating at lower positions.
Why does Google continue to dominate the Chatbot Arena in June 2025?
Google has maintained its dominant position thanks to constant improvements to its API, the development of high-performing models, and strategic partnerships, such as with Canva for video.