Jeremy Feng

AI Research Scientist @Apodex_AI
Learn & Build in Public 👨‍💻

[Simon Shaolei Du @SimonShaoleiDu]

Future prediction is deep research under uncertainty. It requires evidence gathering, tracking change, and careful reasoning. Apodex take the top 4 spots on FutureX with our experimental prediction harnesses — using only Apodex-1.0-mini, our 35B model: https://t.co/jInixUj5Ki

[Apodex @Apodex_AI]

https://t.co/UkIsH0FoSP

[Apodex @Apodex_AI]

Meet 𝗔𝗽𝗼𝗱𝗲𝘅 𝟭.𝟬 🔭 — a heavy-duty agent team for deep research, which sets the SOTA! The team searches the web, reasons over evidence, and writes reports where every claim is backed by an explicit 𝘦𝘷𝘪𝘥𝘦𝘯𝘤𝘦 𝘤𝘩𝘢𝘪𝘯, independently audited before delivery.
🌐 https://t.co/pOQAjL92uF

[Jeremy Feng @JeremyFeng98]

Lucky to have talked with so many cool AI engineers.

Learned a lot from every conversation at the booth — thanks for stopping by!❤️

[MiroMindAI @miromind_ai]

Just wrapped our first @aiDotEngineer singapore. Hats off to @agrimsingh @SherryYanJiang @swyx for this incredible event.
3 days at Capitol Singapore, hundreds of builders in the room.
A year ago, "more AI Engineers than ML engineers" was a forecast. Today, you could feel it on the floor.

At our booth, the most common question wasn't "what does your agent do?" It was "how do you know when it's wrong?" Conversations skipped the demo and went straight to traces, eval setups, error handling, how the agent verifies itself. #AIESG

[Bin Wang @Bin_Wangg]

MiroMind is sponsoring AI Engineer Singapore this weekend.

I’ll be at the booth on Sunday. Stop by, say hi, talk shop. 🚀

[Jeremy Feng @JeremyFeng98]

AI for scientific discovery

[Zonglin Yang ✈️ ICML 2026 @Yang_zy223]

Can we actually TRAIN LLMs for scientific discovery — or only prompt them to brainstorm? 🧬✨

🎉 MOOSE-Star → #ICML2026

Most work on LLMs for hypothesis discovery focuses on inference-time agents or feedback-driven refinement. The core generative process — P(hypothesis | research background), or P(h|b) — has been largely sidestepped: directly training it remains an open problem. We show why: a combinatorial complexity barrier makes naive end-to-end training mathematically intractable.

First scalable recipe for training P(h|b), with clean scaling laws on both training data and test-time compute.

📄 Paper: https://t.co/VH36cDIIz0
💻 GitHub: https://t.co/H3ewP99ydj
🤗 HF: https://t.co/BnLb7tcsRc

🧵👇

[MiroMindAI @miromind_ai]

🚀The MiroThinker API went live!
Interactive scaling —up to 300 tool interactions per task × 256K context:
🧠Two engines:
▸ mirothinker-1-7-deepresearch (235B) — GAIA-Val 82.7 · BrowseComp 74.0 · HLE-Text 42.9
▸ -mini (30B) — BrowseComp-ZH 72.3 (SOTA at 30B)

Plus the agent infra goodies to ship them:
🔌 Disconnect-safe execution — submit / resume / cancel without losing work mid-run
📜Full traces on every run — each step, tool call, and decision logged. SFT / DPO-ready out of the box.

Pre-freeze billing: if our platform fails, you get a full refund.
Cancel mid-run: pay only what compute touched.
From $1.25/M input, 25% OFF at launch.
🔑in the comments👇

[Jeremy Feng @JeremyFeng98]

都用 Rust 重写🌚

[Jarred Sumner @jarredsumner]

Bun v1.3.14 releases tomorrow.

If we do merge the Rust rewrite, this would be the last version in Zig

[Jeremy Feng @JeremyFeng98]

https://t.co/0e494lTHId 真的太有趣&好用了，第一次有了一种当 Manager 的感觉。欢迎各位 Agent 员工！

前几天还刷到 @istdrc 的一期播客 ，非常值得一听～

https://t.co/bY7PSVTIOa

[Jeremy Feng @JeremyFeng98]

@jeffgrimes9 Where can I find the Portfolio tab?

[Jeremy Feng @JeremyFeng98]

@perplexity_ai @AravSrinivas @grok Is this open sourced?

[Alexander Whedon @alex_whedon]

Introducing SubQ - a major breakthrough in LLM intelligence.

It is the first model built on a fully sub-quadratic sparse-attention architecture (SSA),

And the first frontier model with a 12 million token context window which is:

- 52x faster than FlashAttention at 1MM tokens
- Less than 5% the cost of Opus

Transformer-based LLMs waste compute by processing every possible relationship between words (standard attention).

Only a small fraction actually matter.

@subquadratic finds and focuses only on the ones that do.

That's nearly 1,000x less compute and a new way for LLMs to scale.

[Jeremy Feng @JeremyFeng98]

@turingou 赞同👍也许还可以加个审美

[Jeremy Feng @JeremyFeng98]

@0xdeusyu 噢噢，我是 plus，没见过这功能😂谢谢！

[GitHub Projects Community @GithubProjects]

https://t.co/y8fwW4yZjV

[Jeremy Feng @JeremyFeng98]

@miromind_ai 我就用 https://t.co/Fj7f4LleRO 做出行攻略，做完还可以一键分享报告给朋友们看，非常好用👍

[Jeremy Feng @JeremyFeng98]

@elliotchen100 @shanda_group @evermind 公寓大厅有无人超市和自动售货机呀，就在食堂后面😄