Tianhe (Kevin) Yu (@TianheYu) / X

Tianhe (Kevin) Yu

115 posts

Tianhe (Kevin) Yu

@TianheYu

AI Research Scientist @meta MSL TBD. Previously @GoogleDeepMind driving Gemini post-training and RL. Co-created Gemini Deep Think. PhD from @StanfordAILab.

Mountain View, CA

cs.stanford.edu/~tianheyu/

Joined May 2020

Pinned
Tianhe (Kevin) Yu
@TianheYu
Jul 9
Super proud of the release of Muse Spark 1.1 with significant improvement in agentic and coding capabilities! Can’t wait to see how this model can be used in daily development. We are still shipping and the best is yet to come!
Mark Zuckerberg
@finkd
Jul 9
(1) Today we're releasing Muse Spark 1.1 -- a strong agentic and coding model at a very low price. It's available through our new Meta Model API and in Meta AI.
Tianhe (Kevin) Yu
@TianheYu
May 20, 2025
Super proud to have been working on Gemini 2.5 Pro Deep Think with so many impressive SoTA results. Hope it can generate some deep thoughts that are useful for you 🤔
Google DeepMind
@GoogleDeepMind
May 20, 2025
Deep Think in 2.5 Pro has landed. 🤯 It’s a new enhanced reasoning mode using our research in parallel thinking techniques - meaning it explores multiple hypotheses before responding. This enables it to handle incredibly complex math and coding problems more effectively.
Tianhe (Kevin) Yu
@TianheYu
Jun 5, 2025
We improved our Gemini 2.5 Pro model across the board (again)! This time the model will be GA. 🚢🚢🚢
Sundar Pichai
@sundarpichai
Jun 5, 2025
Our latest Gemini 2.5 Pro update is now in preview. It’s better at coding, reasoning, science + math, shows improved performance across key benchmarks (AIDER Polyglot, GPQA, HLE to name a few), and leads @lmarena_ai with a 24pt Elo score jump since the previous version. We also
Tianhe (Kevin) Yu
@TianheYu
Apr 17, 2025
What a smart and efficient model. Gemini 2.5 series keep delivering
Google DeepMind
@GoogleDeepMind
Apr 17, 2025
Gemini 2.5 Flash just dropped. ⚡ As a hybrid reasoning model, you can control how much it ‘thinks’ depending on your 💰 - making it ideal for tasks like building chat apps, extracting data and more. Try an early version in @Google AI Studio → ai.dev
GIF
Tianhe (Kevin) Yu
@TianheYu
Apr 8, 2025
Deep Research + our best model! Time to solve challenging research problems
Google Gemini
@GeminiApp
Apr 8, 2025
📣 Deep Research is now powered by Gemini 2.5 Pro, our most intelligent AI model. ✨ This upgraded Deep Research is now even better at: 🔍 Finding & synthesizing information 📊 Providing more insightful reports 🧠 Analytical reasoning Gemini Advanced users can access the new
Tianhe (Kevin) Yu
@TianheYu
Jul 17, 2025
Such a good model. Generalizes well to the latest IMO problems
Denny Zhou
@denny_zhou
Jul 17, 2025
LLMs for IMO 2025: gemini-2.5-pro (31.55%), o3 high (16.67%), Grok 4 (11.90%). matharena.ai
Tianhe (Kevin) Yu
@TianheYu
May 20, 2025
🙇🤔
Sundar Pichai
@sundarpichai
May 20, 2025
Having a deep think...
Tianhe (Kevin) Yu
@TianheYu
Jul 21, 2025
IMO 🥇achieved! Really proud to have contributed to the post-training and thinking side of this model! Getting closer to ASI!
Google DeepMind
@GoogleDeepMind
Jul 21, 2025
An advanced version of Gemini with Deep Think has officially achieved gold medal-level performance at the International Mathematical Olympiad. 🥇 It solved 5️⃣ out of 6️⃣ exceptionally difficult problems, involving algebra, combinatorics, geometry and number theory. Here’s how 🧵
Tianhe (Kevin) Yu
@TianheYu
May 6, 2025
We just made Gemini 2.5 Pro even better at coding with an improved @lmarena_ai ELO. Hope you’ll like it 🚢
Logan Kilpatrick
@OfficialLoganK
May 6, 2025
Gemini 2.5 Pro just got an upgrade & is now even better at coding, with significant gains in front-end web dev, editing, and transformation. We also fixed a bunch of function calling issues that folks have been reporting, it should now be much more reliable. More details in 🧵
Tianhe (Kevin) Yu
@TianheYu
Mar 25, 2025
And 1443 ELO on @lmarena_ai! So proud to have been working on this great thinking model!
Sundar Pichai
@sundarpichai
Mar 25, 2025
1/ Gemini 2.5 is here, and it’s our most intelligent AI model ever. Our first 2.5 model, Gemini 2.5 Pro Experimental is a state-of-the-art thinking model, leading in a wide range of benchmarks – with impressive improvements in enhanced reasoning and coding and now #1 on
00:00
Tianhe (Kevin) Yu
@TianheYu
Jan 22, 2025
We keep thinking 🤔 and shipping 🚢.
Demis Hassabis
@demishassabis
Jan 21, 2025
Our latest update to our Gemini 2.0 Flash Thinking model (available here: goo.gle/4jsCqZC) scores 73.3% on AIME (math) & 74.2% on GPQA Diamond (science) benchmarks. Thanks for all your feedback, this represents super fast progress from our first release just this past
Tianhe (Kevin) Yu
@TianheYu
Aug 4, 2022
Finally wrapped up my PhD! Super grateful to my wonderful advisor @chelseabfinn all over the years and thanks to @svlevine, @DorsaSadigh, @StefanoErmon & Benjamin Van Roy for being on my committee as well as everyone else that helped me along the way!
Chelsea Finn
@chelseabfinn
Aug 4, 2022
Congratulations to @TianheYu who defended his PhD thesis this week!👏🎓 His work includes: - Meta-World meta-world.github.io - offline model-based RL methods like MOPO and COMBO arxiv.org/abs/2102.08363 - methods for using unlabeled data in offline RL arxiv.org/abs/2202.01741
Tianhe (Kevin) Yu
@TianheYu
Aug 1, 2025
Gemini 2.5 Deep Think is finally out! From winning IMO gold medal to to tackling hard problems in general, this model uses parallel thinking to achieve so. A big step towards superintelligence.
Google DeepMind
@GoogleDeepMind
Aug 1, 2025
For researchers, scientists, and academics tackling hard problems: Gemini 2.5 Deep Think is here. 🤯 It doesn't just answer, it brainstorms using parallel thinking and reinforcement learning techniques. We put it into the hands of mathematicians who explored what it can do ↓
00:00
Tianhe (Kevin) Yu
@TianheYu
Apr 2, 2025
The team has built such a smart model. The gap becomes bigger on harder problems!
Mislav Balunović
@mbalunovic
Apr 2, 2025
Big update to our MathArena USAMO evaluation: Gemini 2.5 Pro, which was released *the same day* as our benchmark, is the first model to achieve non-trivial amount of points (24.4%). The speed of progress is really mind-blowing.