Twitxy Saved

Lucas Liebenwein

Verified account Inco AI

@LLiebenwein

AI, inference, and more to come soon | Former @nvidia, OmniML (acq), PhD at @MIT
New York City lucasliebenwein.com Joined April 2021
295 Following 777 Followers 129 Posts

Latest posts 5

Lucas Liebenwein Verified account @LLiebenwein
Latency and cost decide whether open models work in production. That's the work we do every day at Inco.

Today @zhijianliu is on the KV cache panel at #InferCon, 10:55am in Room B. If you're there, go catch it.
Inco AI Verified account @inco_ai 20h
Infer/Con is today 🧡

10:55am, Room B: our co-founder @zhijianliu is on the "KV Cache as a Distributed Systems Problem" panel at #InferCon, convened by @datadoghq.

At the speaker lunch, he's also hosting a lunch table on speculative decoding and inference for agents. Come join us!
View quoted post on X
Lucas Liebenwein Verified account @LLiebenwein
Your drafter shouldn't be the bottleneck.

@zhijianliu's talk shows how DFlash drafts a whole block in parallel. Lower latency, same output: the target model still checks every token.

Running open models in production and fighting latency or cost? DM me.
Zhijian Liu Verified account @zhijianliu Nov 5
A takeaway from my #Runtime talk recently: drafting one token at a time becomes its own bottleneck.

With #DFlash, we draft a whole block in parallel. That makes room for a deeper, more accurate drafter while keeping latency low. The target model still verifies every token.

At @inco_ai, we’re building on this work to make inference faster for AI applications and agents.
View quoted post on X
Lucas Liebenwein Verified account @LLiebenwein
Latency and cost decide whether open models work in production.

@zhijianliu is speaking at Infer/Con in SF on Nov 16. Go catch his talk.

Can't make it, but fighting latency or cost on open models? DM me.
Inco AI Verified account @inco_ai Nov 4
Inco AI is heading to Infer/Con in San Francisco on Nov 16th, 2026, convened by #Datadog!

Our co-founder @zhijianliu is speaking at #InferCon, alongside researchers and engineers building AI systems and tackling the challenges of inference at scale. Building with open models or working on inference? Let’s connect and swap notes on latency, cost, and scaling AI systems.

Apply at infercon.ai
View quoted post on X
Lucas Liebenwein Verified account @LLiebenwein
Come join us!
Inco AI Verified account @inco_ai Oct 26
Inco is going to #SGLangSummit 2026 🧡

Our open-model endpoints run several times faster than most providers on @ArtificialAnlys. Our co-founder @ZhijianLiu will lead a breakout session on #DFlash and what it takes to build inference for the agentic era, at scale.

🎙️ DFlash and Beyond: How Inco Runs the Fastest Open-Model Endpoints

📍 Nov 12–13, SF
🎟️ luma.com/sglang-summit

Coming to the summit? Stop by, say hi, and let's talk about your inference stack.
View quoted post on X
Lucas Liebenwein Verified account @LLiebenwein
Super excited to bring more speed to local AI!

@inco_ai’s unique approach to pushing the inference frontier from first principles to a shipped product is at full display.

And if you don’t have a powerful local PC you can access our super fast tokens on platform.inco.ai
platform.inco.ai Inco Platform Prepaid, pay-per-token inference. Sign up, add credit, get an API key, and call an OpenAI- and Anthropic-compatible endpoint.
Jian Chen Verified account @jianchen1799 Oct 20
We just shipped Splash 1.4.0!
Qwen3.8-Flash-Next on an M5 Pro (64GB): up to 120 tok/s decode and 1.3K tok/s prefill at 32K.
Over 30% faster prefill and decode than Strata. Up to 1M context, with bigger models fitting in less memory.
github.com/incoai/splash
View quoted post on X