Zhicheng Sun

mts @thinkymachines | prev: @pika_labs, @PKU1898 | generative models, continual learning

[Prophet Arena @ProphetArena]

Getting facts right is memorization. Getting the future right is true intelligence. Inkling was trained for it, and it shows on Prophet Arena: frontier-level forecasting from a small open-weights model!

[Long Lian @LongTonyLian]

应可灵！✨应可灵！✨ bling bling bling✨

[Tim Li @LiTianleli]

Incredibly proud of the team. After countless late nights, Inkling is out, and I especially want to highlight the post-training stack and RL recipes behind it.

A few of my favorite details:

We scaled our largest RL run to 30M+ rollouts and thousands of continuous training steps—with no collapse, no restarts, and stable KL and entropy throughout. Reasoning performance improved log-linearly from the SFT initialization all the way to the released checkpoint. A number of innovations under the hood made this possible, and the result is a strong testament to our post-training technology.

We trained controllable thinking effort directly through RL. By varying the system message and per-token cost, the model learned to trade off tokens and performance on demand.

We also saw an emergent shift in reasoning style: as RL progressed, the chain of thought became increasingly compressed, shedding grammatical overhead. Inkling reasons like a caveman mathematician—a distinctive style unlike that of other open-source models.

We’re also previewing Inkling-small today and plan to release it very soon. It is exceptionally capable for its size, and we expect the community will find it broadly useful.

Building a simple, stable, and scalable RL stack in such a short time was something few thought possible. This team proved otherwise.

[Tinker @tinkerapi]

Inkling is our first open model from @thinkymachines and is now available on Tinker! Check out these quotes from Tinker customers on their experience with Inkling:

@_Mantic_AI: "Not only does Inkling outperform Kimi K2.6 on our forecasting evals, it does so with half the output tokens."

@trajectorylabs: "We’ve been impressed by how sharp and efficient the model is. Its reasoning is concise, its tool calling is consistently strong, and it holds up well on complex, long-horizon agentic tasks. It feels like a meaningful unlock for what teams can build with open-source models designed for customization."

@lightningrodai: "We came away impressed by the model’s underlying reasoning ability. It’s thoughtful, original, and refreshingly unsycophantic.”

[John Schulman @johnschulman2]

Inkling is out today, with open weights and in Tinker. It's been fun to watch this one come together: pretraining began last winter, and starting in mid-January a small team built up the coding, reasoning, and agentic training from there. We learned a lot building it, and I hope people find good uses for it.

[Mira Murati @miramurati]

Our first model, Inkling. Trained from scratch, weights are open, fine-tunable on Tinker today.

[Thinking Machines @thinkymachines]

Today, we are introducing Inkling.

Inkling reasons efficiently across text, image, and audio modalities. We are making the full weights available.

https://t.co/Ghebq5mG30

Available today for fine-tuning on Tinker. Play with it in the Inkling Playground. 🧵

[Mira Murati @miramurati]

Today we share the worldview behind our mission.

Human values don't average out. Local knowledge can't be centralized. The good future has many AIs, raised in different places, shaped by the people they serve, disagreeing with each other the way we do.

https://t.co/A14SurOM2K

[John Schulman @johnschulman2]

We started Thinking Machines a year and a half ago with a couple of instincts: that people should have much more ability to customize models and do research on them, and that even as AI becomes more autonomous, there's a lot more to build to make humans and AIs work well together.

A lot has happened since then, especially the massive progress in agents, so we wanted to revisit those instincts in light of everything we've learned, argue about them, and write down what we actually believe now.

This is where we landed after a lot of debate. I'm happy with it!

[Pika @pika_labs]

Pick up! It’s your AI Self calling 🤳

All Pika AI Self agents can now talk on the phone. For when it’s just too difficult to explain, your thumbs are tired, or you’re craving a more personal connection.

[Cade Li @cadeli1029]

Excited to share the tech report of PikaStream1.0 at @pika_labs!

https://t.co/9KwU2dy40Z

We’re moving beyond just "chatting" with AI—we’re giving agents a true visual identity and a face-to-face interaction experience. Check out our open-source skills to see how we're bringing agents to life.

https://t.co/b56tkmm94B

[Alvaro Cintas @dr_cintas]

Pika just released real-time video chat.

You can send a Google Meet invite to your agent (OpenClaw, Claude, AI Self…) and have them join you on a real-time video call.

Watch my agent book an appointment live for me

[Zhicheng Sun @feifeiobama]

@pika_labs really excited for our new agent skills!

[Pika @pika_labs]

Conversations tend to go better with a face and a voice. That’s why we’re thrilled to release the beta version of the first video chat skill for ANY agent, powered by our new real-time model, PikaStream1.0.

The skill preserves memory and personality, and enables real-time adaptability. And if you use it with your Pika AI Self, they’ll be able to execute agentic tasks during the call 💅

[Pika @pika_labs]

WE INTERRUPT YOUR REGULAR PROGRAMMING to say that Pika AI Selves are officially in public beta!

Everyone can now give birth to an agentic extension of themselves at Pika dot me, or on our new iOS app.

[Pika @pika_labs]

☎️ Hello? AI Selves now have phone numbers! Put them in your imessage or SMS to be there when you’re not, settle arguments in your group chats, and make talking to yourself more normal. More ideas 👇🧵

Plus, we’re letting more people in off of our waitlist! QRT to get your own early access code.

[Zhicheng Sun @feifeiobama]

@OlamicShelter @zephyr_z9 it's 16m instead of 1600m, which is ~120b token
deepseek only uses 150k, which is ~1.2b token

[Pika @pika_labs]

🚨Um, our AI Selves just directed and edited their own documentary series—about working with us.🚨 Meet the Pika AI Self team!

This is the difference between an agent and an AI Self: they aren’t tools; they’re living extensions of your skills, personality, and taste.

Quote retweet this post with a question, and one of our AI Selves might respond.

🎥 🧵👇

[Zhicheng Sun @feifeiobama]

Pika AI Self -> @feifeiobam_

[Pika @pika_labs]

Introducing Pika AI Selves: AI you birth, raise, and set loose to be a living extension of you. They’re rich, multi-faceted beings with persistent memory, and maybe even a peanut allergy. It’s up to you!

Have them send pictures to your group chat. Make a video game about your fish. Call your mom while you do anything but call your mom. The possibilities are as myriad as the stars ✨

Get on the list to give birth to yours at pika dot me

[Zhicheng Sun @feifeiobama]

@pika_labs Pika AI Self -> @feifeiobam_