Google has released Gemini 4 Argon globally as a frontier reasoning model for coding, finance, and law
Google has expanded the availability of the Gemini 4 Argon model from paid developer programs to a global release as a frontier reasoning model for coding, financial analysis, and legal workflows; it continues to offer the robotics variant separately.
Google has released the Gemini 4 Argon model globally following initial safety testing. According to the company, it is a frontier reasoning model focused primarily on complex coding, financial analysis, and legal workflows, available via Google AI Studio, Gemini API, Google Cloud, Search, and the Gemini app. The Gemini 4 Argon for Robotics variant, which handles physical reasoning for robots, remains part of the offering.
The model was originally introduced as a robotics tool: it controls robots, creates interactive 3D training environments from photos, sketches, or videos, and evaluates the quality of training data. In one demonstration, Google generated 77 tasks and more than 4 500 training episodes for a control model from a single photo; according to the company, the success rate of the corresponding task on a real robot rose from 56 to 75 percent, and execution time dropped from 35 to 24 seconds. However, according to the source, the demonstrations require detailed human instructions and guidance.
On the ERQA (Embodied Reasoning Question Answering) benchmark, which tests physical and spatial reasoning in 3D without auxiliary tools, Gemini 4 Argon achieved, according to Google, 87 percent, ahead of the other frontier models (Claude Opus 5.5 82.3%, GPT-6 Astra 81%, Claude Fable 5.1 74%). On the AA-Robotics benchmark from Artificial Analysis, where models directly control a real robotic arm, Argon finished, according to the source, third with 55 points out of 100, behind GPT-6 Astra (69) and GPT-6.1 Sol (65).
Why it matters
The model is no longer limited to robotics and paid developer programs, but is available through standard channels (Search, Gemini app, API), so it will reach a wider range of users and companies working on coding, financial, or legal tasks. At the same time, benchmark results from different sources (ERQA vs. AA-Robotics) show a mixed picture of performance in robotics, which is important for companies considering deployment in physical tasks.
Release card
Gemini 4 Argon
- Inputs
- According to sources, it accepts text, images, sketches, and video as input, and generates text, program code, and interactive 3D simulations for robots as output.
- Availability
- Available globally via Google AI Studio, Gemini API, Google Cloud, Search, and the Gemini app; the robotics variant Gemini 4 Argon for Robotics is offered separately.
- ERQA (Embodied Reasoning Question Answering) 87 % According to Google, this demonstrates the model's ability to solve physical and spatial tasks in a 3D environment without auxiliary tools, with a higher score than the models Claude Opus 5.5 (82.3%), GPT-6 Astra (81%), and Claude Fable 5.1 (74%).
- AA-Robotics (Artificial Analysis) 55 bodů ze 100 In direct control of a real robotic arm across eight tasks, the model finished, according to the source, third, behind the models GPT-6 Astra (69 points) and GPT-6.1 Sol (65 points).
- Complex coding
- Financial analysis and legal workflows
- Generating interactive 3D training environments for robots from photos, sketches, or videos
- According to the source, the robot control demonstrations require detailed human instructions and guidance
- In direct control of a real robotic arm in the AA-Robotics benchmark, the model finished only third, behind two other models
According to the source, the model achieves a higher score on the ERQA benchmark than the compared frontier models Claude Opus 5.5, GPT-6 Astra, and Claude Fable 5.1, while on the AA-Robotics benchmark with direct control of a real robotic arm it finished third, behind the models GPT-6 Astra and GPT-6.1 Sol.
The card summarizes information from the article and any dated corrections, with a link to the original source. It is not our assessment of the model. It does not yet have a dedicated editorial profile. Model selection and other announcements →
What was added since the original report
Verified updates
-
The model has been released globally, accessible via Google AI Studio, Gemini API, Google Cloud, Search, and the Gemini app; Gemini 4 Argon is classified as a frontier reasoning model; Focused on complex coding, financial analysis, and legal workflows in addition to robotics; Specific availability expanded from paid developer programs to global distribution; Robotics is a special version with physical reasoning, not the primary focus
- The model has been released globally, accessible via Google AI Studio, Gemini API, Google Cloud, Search, and the Gemini app
- Gemini 4 Argon is classified as a frontier reasoning model
- Focused on complex coding, financial analysis, and legal workflows in addition to robotics
- Specific availability expanded from paid developer programs to global distribution
- Robotics is a special version with physical reasoning, not the primary focus
Two audiences, two different impacts
What this means
For individuals
Developers and professionals in law or finance gain access to the new frontier reasoning model directly via the Gemini app, Search, or the API, without having to wait for a corporate robotics agreement.
For a business
Companies can now deploy Gemini 4 Argon via Google Cloud, Gemini API, or Google AI Studio for complex coding, financial analysis, and legal work, which extends the model's use beyond its original robotics focus and opens it up to a wider range of enterprise customers.
DevelopmentCheck the original
Event sources
independently confirmed · 2 publishers, 1 independent. We count feeds from the same owner only once.