Pinned Tweet
ThreadWeaver 🧵⚡️ is accepted for an oral presentation at #ICML2026 🎉
ThreadWeaver achieves faster reasoning end-to-end though multiple reasoning agents efficiently working together.
Come to our talk at Thursday, Jul 9, 10 AM and poster at 2:30 PM!
ThreadWeaver achieves faster reasoning end-to-end though multiple reasoning agents efficiently working together.
Come to our talk at Thursday, Jul 9, 10 AM and poster at 2:30 PM!
Our parallel reasoning project ThreadWeaver is now open-sourced 🎉!
Check out our Data Gen/SFT/RL recipe at github.com/facebookresear…
In case you don't know, ThreadWeaver 🧵⚡️ is the first parallel reasoning method to achieve comparable reasoning performance to widely-used sequential long-CoT LLMs, with up to 3x speedup across 6 challenging tasks.
2
8
47
24,503
Long Lian retweeted
We let agent swarms collaborate for 10K hours in simulated agent-owned GitHubs.
Gemini-3.8-flash collaborate the best while gpt-6-astra scores the worst among frontier models tested⁉️
What are the elements of successful agent collaboration at scale? Check World of SWECraft 🧵
Gemini-3.8-flash collaborate the best while gpt-6-astra scores the worst among frontier models tested⁉️
What are the elements of successful agent collaboration at scale? Check World of SWECraft 🧵
15
63
273
30,776
70% price cut is huge. Congrats to the Tinker team! 🎉
Tinkerers have been busy scaling up long-context RL! We’ve made significant improvements to Tinker’s efficiency to support those, and are passing these on with price cuts up to 70%.
GLM-5.3-Flash and DeepSeek-v4.1-Flash are also live for cost-efficient long-context work.
0
3
31
2,215
Long Lian retweeted
We’re growing our evaluation team at @thinkymachines. We care a lot about whether models are actually useful and whether our measurements are good enough to serve as the y-axis for scaling research.
We’ll work across a wide range of problems and support several workstreams, al-bearing internal evals for predictive scaling, turning real user feedbacks and product workflows into evals that close usability gaps, auditing graders and harnesses, and developing new benchmarks for customizability.
If you want to work on this, apply here: jobs.ashbyhq.com/ThinkingMachin…
We’ll work across a wide range of problems and support several workstreams, al-bearing internal evals for predictive scaling, turning real user feedbacks and product workflows into evals that close usability gaps, auditing graders and harnesses, and developing new benchmarks for customizability.
If you want to work on this, apply here: jobs.ashbyhq.com/ThinkingMachin…
19
71
820
86,888
Long Lian retweeted
You can now fine-tune GLM-5.3 with @tinkerapi. It’s the first GLM model on the platform and quite possibly the strongest model available there today.
tinker-docs.thinkingmachines.ai/tinker/models
tinker-docs.thinkingmachines.ai/tinker/models
16
55
909
57,560
Long Lian retweeted
Today, we are launching Tinker grants of up to $50,000 in credits for safety research on open-weight models. We share some project ideas that excite us below; if you’re working on a safety project that could be accelerated by additional Tinker credits, we want to hear from you!
41
132
1,514
280,640
Long Lian retweeted
$ ori [your favorite harness] --model thinkingmachines/inkling:free
(More: openrouter.ai/docs/guides/or….)
(More: openrouter.ai/docs/guides/or….)
We want to improve Inkling’s agentic performance. To help us understand its real-world behavior, we are making it available for free on OpenRouter (only with agentic harnesses) for the next few weeks, starting now. We’ll use the data, disassociated from accounts, to better it.
23
20
560
145,227
Try it out for free and feel free to DM me your feedbacks!
We want to improve Inkling’s agentic performance. To help us understand its real-world behavior, we are making it available for free on OpenRouter (only with agentic harnesses) for the next few weeks, starting now. We’ll use the data, disassociated from accounts, to better it.
1
1
19
1,652
Long Lian retweeted
In-context learning is the holy grail of robot learning. It is challenging because:
1. long-context training and infra (ICL can easily max out disk I/O)
2. need model to follow multimodal (sensorimotor) condition
3. data collection strategy and how to pair data
Tried to get it work in 2024, where we proposed in-context robot transformer, an autoregressive formulation: icrt.dev
Congrats on the launch! Looking forward to seeing few-shot improvement via in-context learning in the future!
x.com/letian_fu/stat…
1. long-context training and infra (ICL can easily max out disk I/O)
2. need model to follow multimodal (sensorimotor) condition
3. data collection strategy and how to pair data
Tried to get it work in 2024, where we proposed in-context robot transformer, an autoregressive formulation: icrt.dev
Congrats on the launch! Looking forward to seeing few-shot improvement via in-context learning in the future!
x.com/letian_fu/stat…
Ever since I started working on robot foundation models a handful of years ago, the broad ability to one-shot in-context learn has been the single most vivid goal in my mind along the long road ahead. It’s something @andyzengineer and I have especially been thinking together on for several years now, and something shared by the whole team at Generalist as a major goal.
But even before anybody started talking about robot foundation models, too, the capability this enables, i.e. generalized one-shot learning, has been both incredibly concrete and elusive. In terminology that I think many other people have come up with as well, in grad school we used to talk about a “Ctrl-C-Ctrl-V” type of capability. See something, and have the robot do it too. This same type of idea inspired the name of the 1970 MIT copy demo (people.csail.mit.edu/bkph/phw_copy_…).
The thing is, it sounds simple, but is incredibly hard since the world is never quite the same when it got copied and where you want to paste it. The real world can be hard to predict and is full of variation. To do this, you need strong generalization, and it needs to be acquired in a single example. Doing this over a wide range of tasks, especially for dexterous tasks, is hard mode.
Lots of the components of the idea of making this all happen have been there for a long time. As an example, this 2017 NeurIPS paper “one-shot imitation learning” arxiv.org/pdf/1703.07326 has excellent vision, with ambition well beyond what was achievable at the time, and although it’s not referred to as “in-context learning” since it was pre-Transformer, it actually uses attention to condition on a single demonstration.
And now, many things have happened since early 2017, including the broadly celebrated arrival of one/few-shot in-context learning in language models in 2020.
This new model GEN-1.5 takes in everything we have built and learned over the past couple years at Generalist. It has been training for 8 months. It has taken an incredible amount of commitment and grit from the whole team to get here. The level to which this model has survived many surgeries has continued to surprise me. And its capabilities have continued to surprise as well. We found compositional generalization on Friday. We found sim2real prompting earlier last week. We filmed the contiguous uncut videos of live prompting yesterday.
To be clear, the success rates are modest, and there’s still a long way to go. But now I have definitely seen a ~decade-long imagination come into the real world.
3
3
101
10,688
Long Lian retweeted
We care about the community and we'd love to learn how you're actually using Inkling!
- What are you building with it?
- Where does it feel strongest or weakest?
- Which multimodal capabilities matter most to you?
- If we made an Inkling Mini/Tiny, what size or sparsity would be genuinely useful?
Real workflows, surprising wins, and failure cases are all helpful. We will award Tinker credits for Good failure cases and prompt contributions! 🫡
And...if you're interested in working on minimal, robust, signal-bearing evals for predictive scaling, and usability evals beyond benchmaxxing for future models, please reach out! 🤔🤖
- What are you building with it?
- Where does it feel strongest or weakest?
- Which multimodal capabilities matter most to you?
- If we made an Inkling Mini/Tiny, what size or sparsity would be genuinely useful?
Real workflows, surprising wins, and failure cases are all helpful. We will award Tinker credits for Good failure cases and prompt contributions! 🫡
And...if you're interested in working on minimal, robust, signal-bearing evals for predictive scaling, and usability evals beyond benchmaxxing for future models, please reach out! 🤔🤖
Models need to be smart and useful. We're building Inkling and Tinker to turn real workflows into better evals, models, and benchmarks---and we want your input!
5
7
39
4,430
Long Lian retweeted
Today, we are releasing Inkling-Small.
Inkling-Small achieves comparable performance to Inkling at a quarter of its size. It features 276B total parameters, 12B active. We are making the full weights available.
thinkingmachines.ai/news/inkling-s…
Fine-tune it on Tinker today, or chat with it in text, image, and audio on Tinker Playground.
Inkling-Small achieves comparable performance to Inkling at a quarter of its size. It features 276B total parameters, 12B active. We are making the full weights available.
thinkingmachines.ai/news/inkling-s…
Fine-tune it on Tinker today, or chat with it in text, image, and audio on Tinker Playground.
165
446
3,349
1,613,786
Long Lian retweeted
Today, we’re introducing [schema]: a harness reaching 99% RHAE with Opus 4.8 + Fable 5 and 95.35% with GPT-5.6 Sol on ARC-AGI-3 Public set.
[schema] makes an LLM think like a physicist. 🧵
[schema] makes an LLM think like a physicist. 🧵
126
292
2,486
969,054
Honored to be in the agentic efforts in Inkling. I felt those sleepless nights were well-spent.
Inkling is just a checkpoint of our progress for the foundation model effort started this January. This is just a preview, and the best is yet to come. ✨
Inkling is just a checkpoint of our progress for the foundation model effort started this January. This is just a preview, and the best is yet to come. ✨
Today, we are introducing Inkling.
Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.
thinkingmachines.ai/news/introduci…
Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵
7
5
102
5,607
Long Lian retweeted
Congrats to @thinkymachines on the release of their open weight model Inkling! We were proud to work with their incredible team on preparing this model for release for the past several months.
Now live on our MCP Atlas and AudioMultiChallenge leaderboards. Inkling tied for 🥇 on rpassing Gemini 3 Pro as the de facto frontier model that supports native audio input.
Also notable, on MCP Atlas Inkling had a low hallucination rate compared to other frontier models.
Now live on our MCP Atlas and AudioMultiChallenge leaderboards. Inkling tied for 🥇 on rpassing Gemini 3 Pro as the de facto frontier model that supports native audio input.
Also notable, on MCP Atlas Inkling had a low hallucination rate compared to other frontier models.
5
24
174
130,541
应可灵！✨应可灵！✨ bling bling bling✨
Today, we are introducing Inkling.
Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.
thinkingmachines.ai/news/introduci…
Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵
21
4
58
5,774
Long Lian retweeted
If you're excited about efficient parallel reasoning 🧠⚡ in LLMs, don't miss the ThreadWeaver paper presentation at #ICML2026.
🗓️Oral: Thu, Jul 9, 2026 • 10:15 AM – 10:30 AM KST
🗓️Poster: Thu, Jul 9, 2026 • 2:30 PM – 4:15 PM KST
📎 threadweaver-parallel.github.io
Feel free to reach out to us if you have any questions (unfortunately I won't be attending in person).
🗓️Oral: Thu, Jul 9, 2026 • 10:15 AM – 10:30 AM KST
🗓️Poster: Thu, Jul 9, 2026 • 2:30 PM – 4:15 PM KST
📎 threadweaver-parallel.github.io
Feel free to reach out to us if you have any questions (unfortunately I won't be attending in person).
ThreadWeaver 🧵⚡️ is accepted for an oral presentation at #ICML2026 🎉
ThreadWeaver achieves faster reasoning end-to-end though multiple reasoning agents efficiently working together.
Come to our talk at Thursday, Jul 9, 10 AM and poster at 2:30 PM!
1
3
16
6,007
A quick recap: ThreadWeaver lets a model coordinate its own reasoning by branching into parallel threads, exploring different solution paths, then joining them back together multiple times during reasoning to reach the answer faster.
1
1
12
1,036
Learn more about ThreadWeaver: threadweaver-parallel.github.io
❤️ Huge thanks to our amazing team members:
@LongTonyLian @sidawxyz @felixudr @tsujuifu @xiuyu_l @YalaTweets @trevordarrell @alsuhr @tydsh @VictoriaLinML
❤️ Huge thanks to our amazing team members:
@LongTonyLian @sidawxyz @felixudr @tsujuifu @xiuyu_l @YalaTweets @trevordarrell @alsuhr @tydsh @VictoriaLinML
0
1
7
918
Long Lian retweeted
We could not be prouder of this years BAIR PhD graduates! Follow their journey to see where they go next @baifeng_shi @sea_snell @databoydg @enfleisig @graceluo_ @zhuhl98 @HaozhiQ @jdzamfi @LianJiachen @joshminwookang @junhaobearxiong @kaylolittlejohn @KentKChang @kvablack @KunheYang @lisabdunlap @LongTonyLian @maulik__bhatt @michaelpsenka @nathanlichtle @neerjathakkar @nikita_mehandru @NiklasLauffer @qiyang_li @vbenara05 @vonekels @xiuyu_l @Yiheng_Li_Cal @ZheFu1 Sampada Deglurkar Wei-Jer Chang Yichen Xie Yigit Erginbas
bair.berkeley.edu/blog/2026/07/0…
bair.berkeley.edu/blog/2026/07/0…
6
16
140
35,887