Gemini 3 5 flash: Comment mieux utiliser l'IA agentique ?

Gemini 3 5 flash: Comment mieux utiliser l'IA agentique ?

Here is the Training: Gemini 3 5 flash! How to better use agentic AI?

🎙 Parlons IA 👥 17K 📅 June 5, 2026 ⏱ 45 min 👁 3K 📄 tutorial 🧭 2026-09-08
Available in: English (current) Français

Keywords

Gemini 3.5 Flashagentic AIprompt engineeringAntigravity CLItoken costs

Summary

This video by Parlons IA offers a practical guide to using Google’s Gemini 3.5 Flash, focusing on optimizing usage in the context of agentic AI. The creator discusses recent changes to Google’s subscription plans, highlighting the shift from unlimited usage to token-based consumption and introducing new Ultra tiers. He provides a workaround for exhausted credits by using multiple Google accounts with the Antigravity CLI tool. The video includes a comparison of Gemini 3.5 Flash with ChatGPT 5.5 and Claude Opus, emphasizing cost differences and performance in tasks like coding and infographic generation. A significant portion is dedicated to prompt engineering, advocating for structured, XML-delimited prompts over vague instructions, and introducing the concept of ‘prompt voodoo’ as an ineffective approach. The creator shares personal tests, noting Gemini’s strengths in coding but weaknesses in image generation and context understanding. He also promotes his own training courses and affiliate links, which adds a commercial layer to the content. Overall, the video serves as a tutorial for users seeking to maximize their use of Gemini 3.5 Flash while managing costs.

176 words

Critical Evaluation

Value of the Information & Strength of the Argument

The video provides valuable, actionable information for users of Gemini 3.5 Flash, particularly regarding subscription management and prompt engineering. The step-by-step guide for installing Antigravity CLI and the tip to use multiple accounts to bypass credit limits are practical and directly applicable. The comparison of token costs across models is useful for budget-conscious users. However, the argumentation is largely based on personal experience and anecdotal evidence rather than systematic testing or official documentation. The creator’s claims about model performance, such as Gemini’s tendency to hallucinate, are presented as general observations without rigorous methodology. The promotional segments for his own courses and affiliate links may bias the recommendations, but the core technical advice appears sound.

Scientific Rigor, Source Quality, Title Accuracy

The video lacks formal citations or references to official documentation, despite mentioning ‘documentation officielle de Google’ in passing. The creator relies on personal testing and experience, which limits the scientific rigor. The title accurately reflects the content, which is a tutorial on using Gemini 3.5 Flash. The video includes promotional content for the creator’s training courses and affiliate links, which is disclosed but may affect objectivity. No viewer comments were provided for analysis.

202 words

Title / Content Match

The title accurately reflects the content, which focuses on practical tips for using Gemini 3.5 Flash, including subscription management, installation of Antigravity CLI, and prompt engineering.

Quality & Reliability

6/10

The video provides practical, hands-on advice on using Gemini 3.5 Flash, including installation steps and prompt engineering techniques. However, it contains promotional content for the creator's own training courses and affiliate links, and some claims (e.g., about model capabilities) are based on personal experience rather than verifiable data. The pricing information is specific but may be time-sensitive.

Key Moments

Cited Sources

Concurring Sources

  • Google AI - Gemini — Official Google page for Gemini, which may confirm model features and pricing.

Dissenting Sources

  • OpenAI - ChatGPT — The video claims ChatGPT 5.5 outperforms Gemini 3.5 Flash in image generation, which may be contested by other reviews.

Contribution & Novelties

The video offers a practical, user-centric perspective on navigating the transition to agentic AI with Gemini 3.5 Flash, particularly focusing on cost management and prompt engineering. It introduces the concept of ‘prompt voodoo’ as a critique of common but ineffective prompting practices, and advocates for structured, XML-based prompts as a more reliable method. The tip to use multiple Google accounts to circumvent credit limits is a novel workaround not commonly discussed. The video also provides a comparative analysis of token costs across major AI models, which is valuable for budget planning.

Pour aller plus loin :

128 words

Radar Profile

The radar profile shows moderate scores across all dimensions, with a slight peak in information quantity (7) and a dip in reliability (5). This suggests the video is informative and practical but lacks rigorous sourcing and may be biased by promotional content.

Reliability 5/10