Ian Ballantyne
@iballantyne.bsky.social
✨AI Developer Relations Engineer @ Google. I work on Gemini API, Gemma, AI Edge and more!
London, UK
https://www.linkedin.com/in/ianballantyne
London, UK
https://www.linkedin.com/in/ianballantyne
How fast is #gemma3 270M out of the box on a @raspberrypi.com 5? About 30 tok/s on CPU for the Q4_0 model using #ollama. Tried using #llamacpp, got about 32 tok/s. 🤯
Extremely promising for edge compute with no accelerator! Model is only ~250Mb on disk. Can't wait to fine-tune for my IoT projects 👨💻
Extremely promising for edge compute with no accelerator! Model is only ~250Mb on disk. Can't wait to fine-tune for my IoT projects 👨💻
August 15, 2025 at 12:39 AM
How fast is #gemma3 270M out of the box on a @raspberrypi.com 5? About 30 tok/s on CPU for the Q4_0 model using #ollama. Tried using #llamacpp, got about 32 tok/s. 🤯
Extremely promising for edge compute with no accelerator! Model is only ~250Mb on disk. Can't wait to fine-tune for my IoT projects 👨💻
Extremely promising for edge compute with no accelerator! Model is only ~250Mb on disk. Can't wait to fine-tune for my IoT projects 👨💻
Reposted by Ian Ballantyne
Great news. We have a new position on the team for a Developer Advocate at the intersection of Web, Chrome, ChromeOS and Android.
The position is in Waterloo - Canada
www.google.com/about/career...
The position is in Waterloo - Canada
www.google.com/about/career...
Senior Developer Advocate, Chrome — Google Careers
www.google.com
March 17, 2025 at 7:12 PM
Great news. We have a new position on the team for a Developer Advocate at the intersection of Web, Chrome, ChromeOS and Android.
The position is in Waterloo - Canada
www.google.com/about/career...
The position is in Waterloo - Canada
www.google.com/about/career...
Understanding an LLM's ability to perform real-world tasks is going to be a much better method of "usefulness". I applaud efforts that attempt to quantify this.
The Agent Leaderboard on Huggingface🔥
Evaluate LLM's ability to utilize tools in complex scenarios. Understand how AI agents perform in real-world business scenarios. Stunning leaderboard created with Gradio 5 😍
Evaluate LLM's ability to utilize tools in complex scenarios. Understand how AI agents perform in real-world business scenarios. Stunning leaderboard created with Gradio 5 😍
February 14, 2025 at 6:50 AM
Understanding an LLM's ability to perform real-world tasks is going to be a much better method of "usefulness". I applaud efforts that attempt to quantify this.
Reposted by Ian Ballantyne
It's been a hot minute since I wrote something on
@hacksterio.bsky.social, but today I put together a quick pro-tip article on connecting an #esp32 #esp8266 to the #Gemini API.
www.hackster.io/PaulTR/conne...
@hacksterio.bsky.social, but today I put together a quick pro-tip article on connecting an #esp32 #esp8266 to the #Gemini API.
www.hackster.io/PaulTR/conne...
Connecting an ESP8266 to Gemini
A quick example of connecting an ESP8266 to the Gemini AI API. By Paul Ruiz.
www.hackster.io
February 11, 2025 at 10:21 PM
It's been a hot minute since I wrote something on
@hacksterio.bsky.social, but today I put together a quick pro-tip article on connecting an #esp32 #esp8266 to the #Gemini API.
www.hackster.io/PaulTR/conne...
@hacksterio.bsky.social, but today I put together a quick pro-tip article on connecting an #esp32 #esp8266 to the #Gemini API.
www.hackster.io/PaulTR/conne...
The price <> capability relationship of Gemini 2.0 models is really exciting right now. Using aistudio.google.com to test the models, then turn those experiments into production apps is much more feasible from a cost basis. ✨💖
Three new Gemini models today: Gemini 2.0 Pro (Experimental), Gemini 2.0 Flash and Gemini 2.0 Flash-Lite
Flash is cheaper than GPT-4o mini, and Flash-Lite is HALF the price of OpenAI's cheapest model!
Updated llm-gemini plugin adding support for the 3 new models: simonwillison.net/2025/Feb/5/g...
Flash is cheaper than GPT-4o mini, and Flash-Lite is HALF the price of OpenAI's cheapest model!
Updated llm-gemini plugin adding support for the 3 new models: simonwillison.net/2025/Feb/5/g...
Gemini 2.0 is now available to everyone
Big new Gemini 2.0 releases today: - **Gemini 2.0 Pro (Experimental)** is Google's "best model yet for coding performance and complex prompts" - currently available as a free preview. - …
simonwillison.net
February 6, 2025 at 10:42 AM
The price <> capability relationship of Gemini 2.0 models is really exciting right now. Using aistudio.google.com to test the models, then turn those experiments into production apps is much more feasible from a cost basis. ✨💖
Just saw a comment saying that model comparisons should only be objective and based on research papers. Completely disagree. Have you ever made a decision based on how you feel? Yeah, thought so.
February 1, 2025 at 5:08 PM
Just saw a comment saying that model comparisons should only be objective and based on research papers. Completely disagree. Have you ever made a decision based on how you feel? Yeah, thought so.
So today's the day I move to Google DeepMind. It's still mind bogging that I'm going to work with the amazing people there to bring life-changing AI research to developers and users. I'm beyond excited to help carve out a future that can really benefit society! 🤩 Let's go! ✨🚀❤️
deepmind.google
deepmind.google
Google DeepMind
Artificial intelligence could be one of humanity’s most useful inventions. We research and build safe artificial intelligence systems. We're committed to solving intelligence, to advance science...
deepmind.google
January 31, 2025 at 11:11 AM
So today's the day I move to Google DeepMind. It's still mind bogging that I'm going to work with the amazing people there to bring life-changing AI research to developers and users. I'm beyond excited to help carve out a future that can really benefit society! 🤩 Let's go! ✨🚀❤️
deepmind.google
deepmind.google
Using LoRA to customize a model like Gemma 2 is a brilliant strategy for model deployment! You get a range of specialist behaviours, while only needing to deploy a single model + N sets of (much smaller) LoRA weights. 🚢=💎+🏋️
Unlock a universe of AI personalities with ONE 💎 Gemma model! 🤯
Customer Service: 💎+❤️ = Empathetic Gemma😊
Marketing: 💎+💡 = Idea Generator Gemma🚀
Coding: 💎+💻 = Code Guru Gemma👩💻
Multiple LoRA adapters on the same GCP endpoint!
Customize your AI and maximize your resources
medium.com/google-cloud...
Customer Service: 💎+❤️ = Empathetic Gemma😊
Marketing: 💎+💡 = Idea Generator Gemma🚀
Coding: 💎+💻 = Code Guru Gemma👩💻
Multiple LoRA adapters on the same GCP endpoint!
Customize your AI and maximize your resources
medium.com/google-cloud...
Open Models on Vertex AI with Hugging Face: Serving multiple LoRA Adapters on Vertex AI
This blog post provides a practical example of how to deploy a Gemma 2 model with multiple LoRA adapters on Vertex AI using custom…
medium.com
January 14, 2025 at 11:24 PM
Using LoRA to customize a model like Gemma 2 is a brilliant strategy for model deployment! You get a range of specialist behaviours, while only needing to deploy a single model + N sets of (much smaller) LoRA weights. 🚢=💎+🏋️
STOP THE PRESS!! It has come to my attention that (some) Costa Express machines at petrol stations now sell Tea. Non-UK folks: This news is as big as a nation collectively winning the superbowl.
January 14, 2025 at 9:51 AM
STOP THE PRESS!! It has come to my attention that (some) Costa Express machines at petrol stations now sell Tea. Non-UK folks: This news is as big as a nation collectively winning the superbowl.
Reposted by Ian Ballantyne
👩💻 Did you know that you can use Google AI Studio's Gemini 2.0 Flash and Gemini 2.0 Thinking Mode models directly in Cursor?
To customize, all you have to do is modify the selected models in your settings:
⚙️ Settings ➡️ Models ➡️ Model Names
To customize, all you have to do is modify the selected models in your settings:
⚙️ Settings ➡️ Models ➡️ Model Names
January 13, 2025 at 4:24 PM
👩💻 Did you know that you can use Google AI Studio's Gemini 2.0 Flash and Gemini 2.0 Thinking Mode models directly in Cursor?
To customize, all you have to do is modify the selected models in your settings:
⚙️ Settings ➡️ Models ➡️ Model Names
To customize, all you have to do is modify the selected models in your settings:
⚙️ Settings ➡️ Models ➡️ Model Names
Reposted by Ian Ballantyne
Wrote up my initial impressions of the new Google Gemini 2.0 Flash model - it's really good, and the streaming mode (where you can stream video and audio to it and get audio streamed right back) is pure science-fiction simonwillison.net/2024/Dec/11/...
Gemini 2.0 Flash: An outstanding multi-modal LLM with a sci-fi streaming mode
Huge announcment from Google this morning: Introducing Gemini 2.0: our new AI model for the agentic era. There’s a ton of stuff in there (including updates on Project Astra and …
simonwillison.net
December 11, 2024 at 8:22 PM
Wrote up my initial impressions of the new Google Gemini 2.0 Flash model - it's really good, and the streaming mode (where you can stream video and audio to it and get audio streamed right back) is pure science-fiction simonwillison.net/2024/Dec/11/...
Reposted by Ian Ballantyne
🎅 Love that developers are already starting to build with Gemini 2.0 Flash!
This project leverages AI + web automation to create an agent capable of navigating and interacting w Instacart The agent is designed to help users efficiently order Christmas-themed groceries.
🔗 github.com/peytoncasper...
This project leverages AI + web automation to create an agent capable of navigating and interacting w Instacart The agent is designed to help users efficiently order Christmas-themed groceries.
🔗 github.com/peytoncasper...
December 20, 2024 at 6:55 PM
🎅 Love that developers are already starting to build with Gemini 2.0 Flash!
This project leverages AI + web automation to create an agent capable of navigating and interacting w Instacart The agent is designed to help users efficiently order Christmas-themed groceries.
🔗 github.com/peytoncasper...
This project leverages AI + web automation to create an agent capable of navigating and interacting w Instacart The agent is designed to help users efficiently order Christmas-themed groceries.
🔗 github.com/peytoncasper...