ML Researcher | Core contributor to Apples official ML framework (MLX)| Violin enthusiast 🎻 | Always coding, sometimes watching Anime.
Germany
Joined February 2022
- Tweets2.3K
- Following887
- Followers3.2K
- Likes69.1K
Pinned Tweet
My name got featured in one of Apples WWDC 2025 session videos 👀🍎
Crazy to think my work made it into a global Apple dev session. I’m proud!!
Thanks @awnihannun @angeloskath @Apple
Gökdeniz Gülmez retweeted
Give me all the GPUs because you don’t know how to use them well 🤣
How convenient…
sam: local inference is a terrible allocation of scarce resources
“a lot of people want local inference, and somehow they assume that the problem is solved now because we have open-source models, we're going to be able to run them on our computers, and so we're not going to need all the stuff in the data centre
but from a global allocation perspective, actually it's quite terrible to have everybody, like, have a bunch of the world's limited supply of DRAM sitting in their own computer and being used five percent of the time
you want maximum utilisation in the cloud if you're trying to deal with this very supply-constrained environment
somehow many people's intuition is the opposite of that”
Gökdeniz Gülmez retweeted
Browser use is one of the biggest unlocks for me personally 🙌🏽
Picking a MacBook is the fun part. Clicking through every option to order it isn't.
Type it once and Nativ does the clicking. It even spots that the touch screen you asked for isn't available.
Running locally on your Mac with @OpenBMB's MiniCPM-V 4.7.
runnativ.ai
Gökdeniz Gülmez retweeted
Coming to MLX-VLM 🚀
Today, we release the weights of a new and improved Le Chonk, advancing the open frontier for agentic coding and cybersecurity.
💻 Built to build: Le Chonk ranks #1 on @cognition’s FrontierCode among open models, and sets a new benchmark for Pareto efficiency on @artificialanalysis.
🛡 Built to defend: State-of-the-art in cyberdefense, put through the paces by the world's leading cyber vendors, including @CrowdStrike, @cloudflare, @rubrik, @zscaler, and others.
Built for sovereignty: Make it yours. Download the weights on @huggingface , customize it with Mistral Forge, and deploy it wherever you need it.
Read more from our team and customers 👇:
Gökdeniz Gülmez retweeted
Some recent awesome optimizations merged in MLX:
Built-in sliding window support in fused SDPA
github.com/ml-explore/mlx/pu…
Use mma/nax for QMV with small rows and large N/K, making batched decoding 2~3x faster
github.com/ml-explore/mlx/pu…
Dequantize fp4 into float16 directly, which makes prefilling 1.4x faster for old macs
github.com/ml-explore/mlx/pu…
Add a 2-stage nax QMM kernel that makes small prompt prefill almost 2x faster for some models
github.com/ml-explore/mlx/pu…
1-bit affine quantization
github.com/ml-explore/mlx/pu…
Thin GEMM kernels making small-M/large-N and large-M/small-N 1.5x faster
github.com/ml-explore/mlx/pu…
1.4x faster fused SDPA by prefetching K/V
github.com/ml-explore/mlx/pu…
And there are actually a lot more awesome changes that I'm too exhausted to list them all, the commit history of MLX is an interesting read for anyone who wants to make open source contributions. github.com/ml-explore/mlx/co…
Gökdeniz Gülmez retweeted
Congrats to @NVIDIAAI on Nemotron 3.5 Transcribe! Day 0 support in Nativ 🎉
40 languages · speaker labels · translation · fully on-device.
On an M3 Ultra at 4-bit:
⚡ a 54 s chat between two people, transcribed with speaker labels in 1.6 s
🚀 27 to 34× faster than real time
💾 about 20 GB at peak, under a third of BF16
🎯 the same transcript as BF16 on every clip we ran
Got a call you'd never upload anywhere? Drop the recording in and get back who said what, without it ever leaving your Mac.
Download for macOS 💻
runnativ.ai
Github repo 🛠
github.com/Blaizzy/nativ
Gökdeniz Gülmez retweeted
Congrats to @OpenBMB on MiniCPM-OCR! Now in Nativ 🎉
OCR 1.0 (1.3B) reads the page in Nativ. Jev (2B), its fact-checker, runs on your Mac too and flags the lines worth a second look.
On an M5 Max in full BF16:
🧾 an invoice read in 0.38 s
📄 a packed full page in under 3 s
🎯 each read checked in about 63 ms
Still retyping totals off scanned invoices? Drop the page in and it's text before you've found the first number, and the scan never leaves your Mac.
Download for macOS 💻
runnativ.ai
Github repo 🛠
github.com/Blaizzy/nativ
Gökdeniz Gülmez retweeted
Excited to have collaborated with @NVIDIAAI to bring Nemotron-3.5-Transcribe to Apple Silicon using @Nativ_AI and our engine 🚀
Meet NVIDIA Nemotron 3.5 Transcribe 🎙️
An open-weight, promptable 30B MoE model with ~4B active parameters that brings transcription in 40 languages, speaker attribution, translation, contextual prompting and audio understanding into one checkpoint.
Watch it keep track of who said what when everyone’s talking at once.
1/ 🧵
Gökdeniz Gülmez retweeted
Earlier this month we met the Docling team at @IBMResearch
Docling turns PDFs, slides and scans into data AI can use. 85M+ downloads.
What they told me: every time they release a model, the first thing Mac developers ask for is an MLX-VLM (by @Neywalabs) version.
That's why Docling has run on MLX-VLM out of the box since 2024. With Granite Docling 258M on an M5 Max, MLX-VLM is 9.2× faster than PyTorch:
→ 1 page: 15.92s vs 1.73s
→ 4 pages: 22.45s vs 2.45s
Thank you @MicheleDolfi, @staar_peter and Max for the chat!
Trying out @liquidai LFM3-Omni on JOSIE-Voice, it’s actually pretty cool and almost has a very uncanny voice. The cool thing is that it has native tool calling so like OpenAI GOT-Live-1 and Moshi RAG you can have synchronized tool delegation. Core backend used is mlx-audio-swift by @Neywalabs still has some bugs though the model sometimes generates only text without the audio, that was the error you see in the end. I used the full precision bf16 model and it was pretty fast on my M4 Pro
Lets keep the releases coming!
Introducing MLX LoRA Studio 3.0.0 🚀🚀🚀
MLX LoRA Studio 3.0.0 brings two ned training algorithms DSLA and KLPO, and finally brings the whole trainer to your favourite agent's fingertips, like @Teknium Herms Agent, @OpenAI Codex @AnthropicAI Claude Code, and al the other that support Skills and MCP with a one click installer.
Whats also new:
- Multiple queued training runs.
- Automatic run cancelation when loss becomes "nan".
- A new live MCP server dashboard
- And more speed and recourse usage optimisations.
Download and try it out:
goekdeniz-guelmez.github.io/…
🚀🚀🚀
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
Gökdeniz Gülmez retweeted
Today we're coming out of stealth, introducing Neywa Labs 🚀
We built the infrastructure to build and deploy AI that runs on the devices people already own. Private, offline, unlimited.
One mission: make AI available to everyone in the world.
neywalabs.ai/
Gökdeniz Gülmez retweeted
Today you will see my true BANKAI 🔥
Stay tuned!
Excited to announce MLX-VLLM 🎉
The first local framework for Vision Large Language inference powered by MLX.
Still WIP and we are open for contributions.
It will be on pipy soon. 🚀
github.com/Blaizzy/mlx-vllm
MLX-VLM's trainer now supports MTP 🚀🚀🚀
github.com/Blaizzy/mlx-vlm/p…
Gökdeniz Gülmez retweeted
Tomorrow is a huge day!
Something we've been building quietly for a while. Stay tuned, the 24h clock starts now 🚀
Finishing up the last two PRs for base VLM training for MLX-VLM by @Prince_Canuma to then start adding decision model training. And then………
Here are the current PRs:
Multi-Token-Prediction: github.com/Blaizzy/mlx-vlm/p…
Quantization Aware Training: github.com/Blaizzy/mlx-vlm/p…