Gemini 3 Pro review – hands-on with Google’s new ‘senior engineer’

Google’s Gemini 3 Pro is a new Artificial Intelligence model focused on advanced reasoning, coding and multimodal analysis; the reviewer tested Antigravity, video analysis and the Deep Think mode to judge whether it can replace senior engineers.

Gemini 3 Pro is presented as Google’s answer to ChatGPT-5 and as an advanced Artificial Intelligence model designed to be clever, concise and more direct than Gemini 2.5. The review focused on three headline capabilities: advanced reasoning, the Antigravity coding interface and multimodal vision that can parse complex PDFs, images and videos. Testing was done with the Cybernews research team and included benchmarks, targeted prompts and hands-on use of Antigravity and the standard Gemini chatbot. Access tiers range from a free tier to paid plans, and the article notes the Google AI Ultra tier is the only plan that unlocks Gemini 3 Deep Think; the review reproduces Google’s pricing labels, including a Free option and a $249.99/month Google AI Ultra reference for deeper access.

On reasoning, Gemini 3 Pro performed strongly in lab tests and the reviewer’s prompts. It scored the highest marks reported on Humanity’s Last Exam and handled a GPQA prompt correctly, answering the color question with “red” and providing a detailed step-by-step explanation. In comparative tests quoted in the article, Gemini 3 Pro recorded “38.3%” on Humanity’s Last Exam, “90.8%” on GPQA Diamond, “74.2%” on SWE Bench and “95.7%” on AIME 2025, outscoring ChatGPT-5 in several categories. Antigravity is a standalone IDE you download for Windows, Mac or Linux; it can create multi-file apps and, after initial server load problems, generated a realistic solar system simulation in around 5 minutes. The reviewer found Antigravity powerful but noted it can be overwhelmed by complex prompts and that generated code still requires human supervision.

Multimodal vision and video analysis were also impressed upon testing. Gemini analyzed complex PDFs without extra prompting, produced sport coaching insights from pickleball and tennis clips, and correctly identified players and tactics in a Steve Nash buzzer-beater. The article flags token costs and limits: Vellum.ai testing cites an input cost of $2.00 per 1 million tokens and an output cost of $12.00 per 1 million tokens for Gemini 3 Pro, and usage allowances vary by tier. The conclusion is that Gemini 3 Pro is an excellent multidisciplinary Artificial Intelligence and a useful tool for advanced users, but it does not replace a senior engineer; it is best used as a force multiplier that still needs a skilled operator to review and secure its outputs.

65

Impact Score

LLM-PIEval: a benchmark for indirect prompt injection attacks in large language models

Large language models have increased interest in Artificial Intelligence and their integration with external tools introduces risks such as direct and indirect prompt injection. LLM-PIEval provides a framework and test set to measure indirect prompt injection risk and the authors release API specifications and prompts to support wider assessment.

NVIDIA may stop bundling memory with gpu kits amid gddr shortage

NVIDIA is reportedly considering supplying only bare silicon to its aic partners rather than the usual gpu and memory kit as gddr shortages constrain fulfillment. The move follows wider industry pressure from soaring dram prices and an impending price increase from AMD of about 10% across its gpu lineup.

Contact Us

Got questions? Use the form to contact us.

Contact Form

Clicking next sends a verification code to your email. After verifying, you can enter your message.