Dmitry Filimonov

Building Agentic AI systems. Co-founder @pyroscopeio.

[Dmitry Filimonov @dmi3f]

@pitdesi safe - simple agreement for founder equity

[Dmitry Filimonov @dmi3f]

I tried all of the TUI coding agents and not a single one of them feels perfect, yet all have some unique features I like.

Maybe the future is everyone building their own.

It's easy to start and when you need new features you just say "hey copy feature X from agent Y".

[Dmitry Filimonov @dmi3f]

There's been so much progress but this still feels like a major AI bottleneck.

[Dmitry Filimonov @dmi3f]

I wonder what it would like to automate the agency part too. I know people are working on AI agents but ironically it doesn’t feel like agents have agency. You still need to tell them what the goal is.

[Dmitry Filimonov @dmi3f]

And over time the tools are getting simpler and simpler. It's the RISC-ification of LLM agents.

[Dmitry Filimonov @dmi3f]

I've been playing with tool_use a lot lately and keep leaning more towards a similar approach in my code — define some tools the system can use, and the business logic becomes just a prompt.

[Dmitry Filimonov @dmi3f]

Assuming it's legally clean, if they have a model that is just as capable but much faster and cheaper that's more or less all you need to justify a valuation.

[Aakash Gupta @aakashgupta]

Cursor is raising at a $50 billion valuation on the claim that its “in-house models generate more code than almost any other LLMs in the world.” Less than 24 hours after launching Composer 2, a developer found the model ID in the API response: kimi-k2p5-rl-0317-s515-fast.

That’s Moonshot AI’s Kimi K2.5 with reinforcement learning appended. A developer named Fynn was testing Cursor’s OpenAI-compatible base URL when the identifier leaked through the response headers. Moonshot’s head of pretraining, Yulun Du, confirmed on X that the tokenizer is identical to Kimi’s and questioned Cursor’s license compliance. Two other Moonshot employees posted confirmations. All three posts have since been deleted.

This is the second time. When Cursor launched Composer 1 in October 2025, users across multiple countries reported the model spontaneously switching its inner monologue to Chinese mid-session. Kenneth Auchenberg, a partner at Alley Corp, posted a screenshot calling it a smoking gun. KR-Asia and 36Kr confirmed both Cursor and Windsurf were running fine-tuned Chinese open-weight models underneath. Cursor never disclosed what Composer 1 was built on. They shipped Composer 1.5 in February and moved on.

The pattern: take a Chinese open-weight model, run RL on coding tasks, ship it as a proprietary breakthrough, publish a cost-performance chart comparing yourself against Opus 4.6 and GPT-5.4 without disclosing that your base model was free, then raise another round.

That chart from the Composer 2 announcement deserves its own paragraph. Cursor plotted Composer 2 against frontier models on a price-vs-quality axis to argue they’d hit a superior tradeoff. What the chart doesn’t show is that Anthropic and OpenAI trained their models from scratch. Cursor took an open-weight model that Moonshot spent hundreds of millions developing, ran RL on top, and presented the output as evidence of in-house research. That’s margin arbitrage on someone else’s R&D dressed up as a benchmark slide.

The license makes this more than an attribution oversight. Kimi K2.5 ships under a Modified MIT License with one clause designed for exactly this scenario: if your product exceeds $20 million in monthly revenue, you must prominently display “Kimi K2.5” on the user interface. Cursor’s ARR crossed $2 billion in February. That’s roughly $167 million per month, 8x the threshold. The clause covers derivative works explicitly.

Cursor is valued at $29.3 billion and raising at $50 billion. Moonshot’s last reported valuation was $4.3 billion. The company worth 12x more took the smaller company’s model and shipped it as proprietary technology to justify a valuation built on the frontier lab narrative.

Three Composer releases in five months. Composer 1 caught speaking Chinese. Composer 2 caught with a Kimi model ID in the API. A P0 incident this year. And a benchmark chart that compares an RL fine-tune against models requiring billions in training compute without disclosing the base was free.

The question for investors in the $50 billion round: what exactly are you buying? A VS Code fork with strong distribution, or a frontier research lab? The model ID in the API answers that.

If Moonshot doesn’t enforce this license against a company generating $2 billion annually from a derivative of their model, the attribution clause becomes decoration for every future open-weight release. Every AI lab watching this is running the same math: why open-source your model if companies with better distribution can strip attribution, call it proprietary, and raise at 12x your valuation?

kimi-k2p5-rl-0317-s515-fast is the most expensive model ID leak in the history of AI licensing.

[Dmitry Filimonov @dmi3f]

The most surprising thing to me is that usually it's the bigger company (by headcount) that runs into this issue, but here it's the other way around.

[Dmitry Filimonov @dmi3f]

@bcherny pls make it so that i can talk to it the same way as i talk to chatgpt voice

[Dmitry Filimonov @dmi3f]

@clairevo 100% — lack of sleep messes with the rational parts of the brain. It was a big “aha” moment for me when the baby started sleeping through the night.

[Dmitry Filimonov @dmi3f]

@mcuban labs likely run models at high margins. open-source models could lower costs short term. running locally means paying mainly for hardware and electricity.

but i guess if everyone does that, hardware and power costs rise so adoption won’t be instant. but long term agents will win

[Dmitry Filimonov @dmi3f]

always has been

[Max Rozen @RozenMD]

wild how a fast CI/CD pipeline with good test coverage is basically a competitive advantage now

[Dmitry Filimonov @dmi3f]

@owengretzinger haha, saw your previous tweet and started building the same thing. I love how quickly you can just build things these days

[Dmitry Filimonov @dmi3f]

And conversely, it was designed by only about five people.

[Anna Riedl @AnnaLeptikon]

Somehow it was learning how many people are fulltime employed to maintain the Golden Gate Bridge that flipped something inside of me in my understanding of the entropic force civilization has to constantly fight against. Before that moment I thought — I had not applied real conscious thought — you simply build a building or anything really and then you just … have it. After that I understood everything is constantly at the brink of being lost.

[Dmitry Filimonov @dmi3f]

@scottastevenson @JoshConstine @nikitabier i’m convinced at this point that this bug increases engagement that’s why it’s never fixed

[Dmitry Filimonov @dmi3f]

@alexeykozy I’ve found that for performance or memory leak issues, connecting Cursor to an MCP server with profiling data gives really good results.

[Dmitry Filimonov @dmi3f]

I wish our industry did 1 week trials instead of 8-round interviews. Part ways quickly if it’s not a fit. It’s a lot harder to fake being good at the actual job.

[Suhail @Suhail]

PSA: there’s a guy named Soham Parekh (in India) who works at 3-4 startups at the same time. He’s been preying on YC companies and more. Beware.

I fired this guy in his first week and told him to stop lying / scamming people. He hasn’t stopped a year later. No more excuses.

[Dmitry Filimonov @dmi3f]

@Sirupsen A lot of database-on-top-of-S3 solutions are essentially adding a buffering layer that holds recent data, ensures durability, and flushes it to S3 in larger, consistent chunks. With conditional writes, much of the complexity can go away — really cool to see you all using it.

[tiffany jernigan 🍃 @tiffanyfayj]

Hey folks, on Thursday July 3rd at 8a PT/11 ET/5p CEST is the @PyroscopeIO community call. Bryan Huhta is going to talk about the #Pyroscope tools in the @Grafana MCP server and some of what’s new since the last call. Come join us!

https://t.co/bcucZIigeA

[Dmitry Filimonov @dmi3f]

my favorite new cursor rule

[Dmitry Filimonov @dmi3f]

so we got iOS 7 today

[Dmitry Filimonov @dmi3f]

Writing the best code of my career but dId YoU kNoW tHe MoDeLs aReNt AcTuAlLy ReAsOnInG??