AI Education @FireworksAI_HQ 🔥🎆 NLP + Gen AI Expert / LLM whisperer 🤖 AI Author 📚 Founder @Aikylie (acquired) 📈 Fellow @YCombinator

San Francisco, CA
I'm especially looking at "mixed across multiple harnesses." Harnesses change pretty often. Teaching a model to adapt across tasks AND harnesses at this scale will really help future proof our models. Not a new technique, but cool to see it become more and more the default.
Nearly half a year of silence. We spent it studying one problem: how far RL can scale. MiMo-V2.6 is in the middle of its RL run right now. Three things we scaled: compute (~2B tokens per step, 1568 prompts × 16 rollouts, fully async), environments and harnesses (multi-task agentic RL, mixed across multiple harnesses in one run), and grader compute (agentic in-group credit assignment, with test-case and rubric-based rewards). We'll open-source the details piece by piece over the coming weeks. Streaming the run: mimo.xiaomi.com/rl/
1
2
282
Companies adopting AI walk the same path: - Rent closed frontier models - Engineer harnesses around them - Move to open models for performance & efficiency - Train logic & skills into weights @harvey, @cursor_ai, @cognition know this. AI-native companies just walk it faster.
2
7
695
I gotta try this
Introducing Atlas: The world's first multimodal world model that generates image and video frames with pixel-perfect camera control and reconstructs them in 3D. Model the world, move the camera, and simulate space & time.
3
210
Something I wish I got asked a lot: "What reasoning level should I use?" Let's try on a computer use dataset: - NO model did best at the max reasoning - K3 and GPT 5.5 really overthink at max - Opus did best at low Test reasoning efforts, please! data + code below
1
2
179
Congratulations on being one of our winners! I took a look and was very impressed :) I'm curious where you got the inspiration for this!
#KimiK3onFireworks I built MedRecap, a home medication reconciliation assistant, with Kimi K3 on @FireworksAI_HQ. Upload photos of your pill bottles, prescriptions, and visit notes, and it organizes everything into one medication list, flags discrepancies across the records, and uses FDA drug labels to surface questions worth confirming with a doctor or pharmacist. All demo data is fictional. 👇 I originally planned to build an outdoor hiking guide. Then I discovered a mature product already doing it well, so I switched ideas close to the deadline. The core of MedRecap was built by K3 through Cursor IDE Agent, covering the frontend, backend, image understanding, and report workflow. I spent about $20 on API usage. What impressed me most is how capable open-source models have become at delivering a complete web app, and how stable the @FireworksAI_HQ API was across vision tasks and multi-turn workflows.
1
1
5
6,197
Nvidia buying Huggingface is going to be a huge shift for local models and open models in general. I’m still wrapping my head around how the world’s largest repository of models and datasets might change because of this.
1
4
155
Sinan Ozdemir retweeted
Check out this month's AI Systems Performance Engineering meetup: "Deep dive into Fireworks AI continuously improving their RL post-training through Rollout, Reward, Update, and Repeat!" by @Prof_OZ @FireworksAI_HQ !! with my co-organizer @anbarth piped.video/s1j7-x6uk_0
1
1
5
596
This was a fun one! When I'm looking at these entries I'm looking at a few things like: 1. Can I try it? 2. Is this solving a real problem? How big of a problem? 3. Does the app goes beyond "just ask the AI"? 4. How ambitious was it? Excited to go through more!
Thanks to everyone that submitted to the Fire Pass Kimi K3 Community Build Contest! @Prof_OZ our Head of AI Developer Education is testing all of the entries, and highlighting the projects and the builders! First up: FridgeChef by @seanchenqt
3
208
Here at #MongoDBlocal Build Fest with @mongodb and @OpenRouter and others working with some amazing folks building on @FireworksAI_HQ
1
10
1,020
Probing will always be one of my favorite parts of LLM/deep learning research 🤩
We tested out Anthropic’s J-Lens probe at two open models: Kimi K3 and Qwen 3.5-9B. By looking at their internal states, we found that the models are setting up the right vocabulary before a single token hits the page. Check out the research: fireworks.ai/blog/J-Lens-Kim…
4
190
This is going to be one of my favorite talks I can already tell! I've been helping companies move from AI dependency to being AI-native for over a decade and I can't wait to keep spreading that good word.
Most companies still rent AI by the token, build on someone else’s roadmap, and hope the next model release does not disrupt their systems. At ODSC AI West 2026, @Prof_OZ, Head of AI Developer Education at @FireworksAI_HQ and AI & LLM Advisor at Tola Capital, will deliver the keynote “Own Your Intelligence: The Path from AI Dependent to AI Native.” 👉 Register now: hubs.li/Q04rCrlC0
3
242
What a week for open models 😍 Sorry weekend plans, gotta test some stuff
🚀 DeepSeek-V4-Flash Official API is now LIVE in public beta! 🔷 We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview. Check out the massive performance leap below! 👇 🔷 The official V4-Flash now natively supports the Responses API format and is fully adapted for Codex! Check out the configuration details in our official API docs: api-docs.deepseek.com/quick_…
1
304
Period.
If LoRA is underperforming, don't reach for more expensive full parameter fine-tuning right away. We ran three cheap tests (data coverage, optimization, rank) to see if we could close the gap between LoRA and FullFT. Sometimes we could, sometimes we couldn't. fireworks.ai/blog/three-test…
2
2
9
2,251
Let me know what questions you all have so we can get them answered for you live! I hope you're excited to see some agents on K3 getting better through Reinforcement Learning, I know I am 🤩
K3 is here. The world's first open frontier model in the 3-trillion-parameter class. Come see how to run it, train it, and put it to work, on Fireworks. Hosted by @jamesr66a and @Prof_OZ on Tuesday, August 4th, 10am PDT. Register here: fireworks.ai/event/Webinar-K…
1
3
479
What are you going to build? 🫵
Want to win a Fire Pass? You have one week to build something that changes your daily life using Kimi K3 on Fireworks. To enter: - Tweet a demo video - Tag us and use #KimiK3onFireworks in your Tweet - Submit your entry here: forms.gle/yZsJWkecYjPVodiK6 5 winners get a unique Kimi K3 Fire Pass.
2
210
Inference + Training + K3 = 🔥🔥🔥 Seriously this is game changing. Using the worlds best open models is one thing, but to also have access to training on top of them day 0 as well? The world is your oyster now go find that pearl
Kimi K3 is live on Fireworks. Day 0, inference and training. US-hosted, and zero data retention. This is the first frontier open model in the 3 trillion parameter class. It sports 1M context, native vision, and reasoning that rivals the top closed models. Boom.
1
1
6
469
This is a big deal to put it mildly.. I genuinly can't wait to see what everyone builds with Kimi K3. Show me what you're working on!
Kimi K3 is now live on Fireworks. Get started → fireworks.ai/models/firework…
1
6
499
Judging last week's @daytonaio HackSprint (w/ @FireworksAI_HQ @braintrust @CopilotKit @coderabbitai ) was a blast! The winner was SubStrata, a land-cover classifier that self-improves. Fine-tuning + inference on Fireworks = 🔥 👏 @AnishG_07, Shivansh Bansal & Neev Tamboli 👏
3
3
13
666