Gemini 3.6 Flash vs 3.5 Flash: The Benchmarks Are Real, The Prompting Gap Is Bigger

From the creator

Link to our newsletter: https://bitbiased.ai/ Gemini 3.6 Flash just replaced 3.5 Flash as Google's default fast-tier model — and everyone's covering the launch, almost nobody's covering what actually determines whether you notice the upgrade: how you prompt it. In this video, I break down two things: 1️⃣ The real benchmark comparison — what Google's own DeepMind numbers show on DeepSWE, OSWorld-Verified, and GDPval-AA, versus what Artificial Analysis's independent Intelligence Index actually confirms. Spoiler: the story is more nuanced than the launch headlines. 2️⃣ A practical prompting guide built specifically around what changed in this model — multi-step staging, context-rich debugging prompts, built-in self-checks, and knowing when NOT to over-structure a simple ask. Covered in this video: What's confirmed vs. what's still just a Google claim Gemini 3.6 Flash pricing vs 3.5 Flash (input/output token costs) Real before/after prompt examples for coding and agentic tasks Why the Artificial Analysis Intelligence Index score didn't move Who should actually test this model this week — and who shouldn't bother yet How Flash Cyber and 3.5 Flash-Lite fit into this same release (full breakdown coming next video) If you're building on Gemini for real production work, this is the 12 minutes that decides whether you get better output tonight or leave performance on the table for weeks. 🔔 Subscribe for the next video, going deep on Flash Cyber, the restricted government-only model from this same release. 💬 Drop a comment: have you run 3.6 Flash against 3.5 Flash on your own workload yet? Does the gap match Google's claims? Timestamps: 00:00:00 Intro 00:01:27 The Upgrade, Stripped Of Launch-Day Noise 00:02:30 Benchmarks Vs. 3.5 Flash: What's Confirmed And What's Still A Claim 00:06:19 The Real Story: Why The Prompting Gap Is Bigger Than The Model Gap 00:07:21 How To Actually Prompt Gemini 3.6 Flash For Better Results 00:11:09 Who Should Actually Test This Today 00:12:19 What Nobody's Putting In The Thumbnail 00:13:30 The Verdict #Gemini36Flash #GeminiAI #GoogleAI #PromptEngineering Tags

Stop watching · start building
Matched to Gemini AI

Makers residency 4th edition

Your idea can be anything. At our last workshops, people built plant-based nutrition businesses, a way to measure planetary resilience, crochet headwear and a way to connect lonely people living in the same building. An asset manager went on to raise an additional £25M on their fund within nine months. The fourth workshop is a full day at KOKO with Nick Sarafa, learning to build with Claude Code. Bring the thing you keep meaning to start. We’ll bring the pizza.

◆ Fri 13 Nov 2026 ◆ KOKO, London
Makers residency 4th edition
Live event
Makers residency 4th edition
Fri 13 Nov 2026

More like this

Running one yourself?

List your AI event,
wherever it is.

A meetup, a workshop, a hackathon, a conference. Any city, or online. Tell us about it and it lands in front of people already learning this stuff.