In partnership with
In Today’s Issue:
⚡ Gemini 3 Flash is here, bringing frontier speed to production.
🌐 Firefox plans to transform into a "modern AI browser" by 2026
💰 OpenAI is reportedly in talks for a $10B+ investment from Amazon
📱 The ChatGPT App Directory launches alongside an Apps SDK
✨ And more AI goodness…
Dear Readers,
Gemini 3 Flash just turned “frontier” performance into something that feels… casual: faster loops, lower latency, and benchmark numbers that basically dare builders to ship agents in production instead of babysitting prototypes. In today’s issue, we zoom in on what this speed-first shift means for your stack (cost, caching, batch workflows, and why “good enough” just leveled up), then swing through the rest of the day’s tension points: Firefox promising an optional “AI Mode” while its privacy-minded base recoils, OpenAI reportedly flirting with $10B+ from Amazon and AWS Trainium as the plot twist in the compute wars, and the ChatGPT App Directory pushing the chatbot toward a real platform economy. Plus: a genuinely wild biology headline—frog gut microbes wiping tumors in mice—and a fresh podcast appearance from Kim + Peter on superintelligence. Ready to connect the dots before everyone else does?
All the best,
Kim Isenberg
Firefox will evolve into an AI browser
Mozilla’s new CEO says Firefox will “evolve into a modern AI browser” over the next ~3 years, with an optional “AI Mode” planned for 2026 that lets users pick between different models (including open-source / Mozilla-hosted options). The internet reaction has been loud and negative, with many Firefox loyalists warning this feels like trend-chasing and a potential privacy/UX regression—even if the AI features can be turned off.
OpenAI: $10b in Amazon Chips
OpenAI is reportedly in talks for $10B+ from Amazon, at a valuation exceeding $ 500B. The kicker is strategic: OpenAI would start using AWS Trainium, giving Amazon a flagship “frontier” customer while OpenAI diversifies away from its Nvidia-heavy stack.
The ChatGPT app store is here
OpenAI just launched a ChatGPT App Directory plus an Apps SDK, pushing ChatGPT toward a real platform where developers can build interactive experiences that run directly inside the chatbot UI.
They also renamed “connectors” to “apps” (e.g., apps with file search/deep research/sync), expanded integrations like Spotify/Zillow. They added new ones like Apple Music and DoorDash, while only vaguely hinting at future monetization (e.g., digital goods).
Co-founders Kim and Peter are guests on a German AI podcast discussing superintelligence and the future of AI.
Gemini 3 Flash Goes Turbo
The Takeaway
👉 Google released Gemini 3 Flash, positioning it as a fast, lower-latency “frontier” model for everyday building and production use.
👉 Google reports strong benchmark results, including 90.4% on GPQA Diamond, 33.7% on Humanity’s Last Exam (no tools), and 81.2% on MMMU Pro for multimodal reasoning.
👉 For coding, Google cites 78% on SWE-bench Verified, framing Flash as a capable agentic coding model.
👉 Google emphasizes cost + speed: pricing starts at $0.50 per 1M input tokens and $3 per 1M output tokens (audio input $1 per 1M), plus context caching (up to 90% savings) and a Batch API (50% savings), with availability across Gemini app, AI Studio/Vertex AI, and developer tooling.
Google just dropped Gemini 3 Flash, a “frontier” model tuned for speed - and the numbers suggest this isn’t just a minor point release. On tough academic reasoning benchmarks, Google reports 90.4% on GPQA Diamond and 33.7% on Humanity’s Last Exam (without tools), plus 81.2% on MMMU Pro for multimodal reasoning. The point: you get Pro-grade thinking without Pro-grade waiting.
For builders, Flash is positioned as the new default workhorse: 78% on SWE-bench Verified for agentic coding, and Google claims it’s ~3× faster than Gemini 2.5 Pro while costing a fraction. Pricing in the Gemini API/Vertex AI starts at $0.50 per 1M input tokens and $3 per 1M output tokens (audio input $1 per 1M). Add context caching (up to 90% savings on repeated prompts) and a Batch API option (50% cost savings), and “iterate fast” suddenly becomes a budget line item, not a vibe.
It’s rolling out in the Gemini app and AI Mode in Search, plus Google AI Studio, Gemini CLI, Antigravity, Android Studio, and Vertex AI.
Why it matters: Fast, cheaper “frontier-ish” reasoning moves the bottleneck from model quality to product design - more teams can afford agents and multimodal workflows in production. If Flash holds up in your evals, the baseline for “good enough” assistants just jumped.
Sources:
🔗 https://blog.google/products/gemini/gemini-3-flash/
🔗 https://deepmind.google/models/gemini/flash/
Turn AI Into Your Income Stream
The AI economy is booming, and smart entrepreneurs are already profiting. Subscribe to Mindstream and get instant access to 200+ proven strategies to monetize AI tools like ChatGPT, Midjourney, and more. From content creation to automation services, discover actionable ways to build your AI-powered income. No coding required, just practical strategies that work.
Frogs against cancer
Researchers in Japan screened 45 bacterial strains taken from the intestines of amphibians and reptiles (tree frogs, fire-belly newts, and grass lizards). Nine strains showed anti-tumor effects, but one was the knockout: Ewingella americana. In a mouse colorectal cancer model, one dose produced a 100% “complete response,” beating both an anti-PD-L1 checkpoint antibody and doxorubicin.
The trick is a two-part “living drug.” First, the bacterium homes in on the tumor’s low-oxygen core and multiplies roughly 3,000× within 24 hours, delivering its toxins where they hurt cancer most. Second, its presence sounds an immune alarm, flooding the tumor with neutrophils plus T and B cells and ramping up inflammatory signals. Early safety checks were encouraging: fast blood clearance, no long-term organ colonization, and short-lived inflammation — with antibiotics as a potential off-switch.
If AI can help us map and screen nature’s microbial toolbox at scale, how many more one-shot therapies are hiding in plain sight?
