fatih c. akyon

AI @ultralytics, PhD @metu_odtu.
author of YOLO26/YOLO27, UltraViT and SAHI.
20M+ open-source downloads, 5 patents, 2500+ citations.

[fatih c. akyon @fcakyon]

I have been working on this for a very long time, finally it's live on Arxiv!

Introducing the long-awaited @Ultralytics YOLO26 paper: state-of-the-art real-time object detection, instance segmentation, image classification, keypoint detection, oriented bounding boxes detection and open-vocabulary detection models that run on any edge device 🔥

Meet me at @CVPR booth #513 if you have questions about the paper!

It was a lot of fun collaborating with my coauthors @arttestt and others.

Of course, it wouldn't be possible without the ultralytics package maintainers like @onuralpszr and rest of the team.

[fatih c. akyon @fcakyon]

@VZHydra

[Jack Lu ✈️ NeurIPS 26 @Jacklu_me]

Even the simplest weight edit, a deletion, can teach an LLM new skills!

Meet Deletion Tuning (#NeurIPS2026): just zeroing <0.1% of weights matches full fine-tuning's accuracy at scale. It's thousands of times smaller and avoids forgetting, a step toward continual learning. 🧵👇

[fatih c. akyon @fcakyon]

Another jev killer, this time by google

[Sundar Pichai @sundarpichai]

Today we’re launching the Gemini Decisions API, powered by our new Gemini Decision 1 model in @GoogleAIStudio.

Instead of generating text token by token, it’s built for closed-set choices scoring up to 1,000 questions across text, images, audio, video + PDFs in a single pass. A few details:

- 3 decision types: yes/no, multiple choice (up to 255 options) + rubric scoring
- $0.10 / 1M input tokens with $0 billed for output tokens (~5x cheaper than Flash-Lite)
- leads on several benchmark evals with lowest calibration error

Google teams are already building with it, below is a look at Stitch Rapid designing a full product flow in seconds (1x speed)

[Ultralytics @ultralytics]

Claude Fable 5.2 refused every baggage X-ray in our benchmark, even a scan of an ordinary phone. 🚫

Every open-weights and Chinese model we tested answered all 10.

A closed lab censors. Open weights answer.

See the refusals ➡️ https://t.co/gBkMKFm11z

#Ultralytics #OpenSource #AI

[fatih c. akyon @fcakyon]

@tokufxug as the maintainers of sahi (me and @onuralpszr), we are glad you liked it 💯

[fatih c. akyon @fcakyon]

@VZHydra you cannot, there is so much to do 😎

[Seth Karten @sethkarten]

always start with evals. work from first principles

[Yavuz Kömeçoğlu @YavuzKomecoglu]

🎙️ @ThinkVoiceAI olarak kendi geliştirdiğimiz Türkçe TTS modellerimizden yeni bir demo!

Videoda, yapay zeka asistanımız iki kişi arasındaki telekonferansı gerçek zamanlı takip ediyor, gerektiğinde konuşmaya dahil olup görüşmeyi yönetebiliyor.

Duyduğunuz üç ses de kendi eğittiğimiz Türkçe TTS modellerimize ait. Üstelik STT, TTS ve LLM dahil tüm altyapımız, herhangi bir dış servise bağımlı olmadan tamamen kendi sunucularımızda çalışıyor.

Bu teknolojiyi kurumlara Private Cloud, On-Premise veya tamamen offline olarak sunabiliyoruz.

Araştırmadan ürünleştirmeye kadar emeği geçen ThinkVoice Ar-Ge ekibimize teşekkürler! 👏

Farklı sesler ve senaryolarla yeni örnekler paylaşmaya devam edeceğiz! 🚀

[fatih c. akyon @fcakyon]

@VZHydra Atm discussing with my agents on how I can use mindware+openship agents to automatize maintenance, thanks for releasing new features every week 💯

[Daniel Smolsky @DanielSmolsky]

@claudeai subscriptions in @opencode are back! Check it out

[fatih c. akyon @fcakyon]

@stoic_tiger @googlegemma @googledevs “AI Edge Foresight” is the worst product naming ever

[fatih c. akyon @fcakyon]

how 🤯

[fatih c. akyon @fcakyon]

AUROC 0.96 model's person/vehicle detection results 🤔

[Xunzhuo @XunzhuoLiu]

📦 New: Box. Ask where something is, get one box with a calibrated confidence.

Four questions about one photo, one request: the frisbee 0.99, the dog's tail 0.98, the cone 1.00.

0.90 accuracy at IoU 0.5 · confidence AUROC 0.96

[DeepSeek @deepseek_ai]

🚀 DeepSeek-V4.1-Flash-1017 is here!

Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.

🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.

1/4

[Ultralytics @ultralytics]

@Google's Gemini 4 Argon is now available in Ultralytics Platform! 🚀

In our latest internal benchmark of 26 AI models across 36 real-world object detection datasets, Gemini 4 Argon ranked #1 in auto-annotation accuracy with 0.565 mAP50, outperforming Claude Opus 5.5 and GPT-6 Astra.

From medical imaging to retail, traffic & industrial inspection, developers can use Gemini 4 Argon to generate bounding boxes, review predictions & annotate datasets at scale, reducing manual labeling effort.

Read more ➡️ https://t.co/gKWNhOoJA4

#Gemini4 #Ultralytics #AI

[fatih c. akyon @fcakyon]

@openshipio Have been waiting for this 🔥

[fatih c. akyon @fcakyon]

@VZHydra Claude code

[Onuralp S. @onuralpszr]

I'm using and contributing @fedora for over 15 years. Always give back.💙

I got a new Dell XPS 16 and saw how @OmarchyLinux, Intel and Dell brought Panther Lake to Linux so beautifully, I wanted to bring that care to @fedora with @kdecommunity Plasma.

https://t.co/sFNAnUURDS

[Google Gemma @googlegemma]

Introducing EmbeddingGemma 2! 🚀

Our lightweight, multimodal embedding model maps text, code, images, video, and audio into a single, unified embedding space. Optimized for on-device use cases, it features:

- 740M parameter form factor with modular encoders
- Flexible dimension sizes (768dim-128dim) via Matryoshka Representation Learning (MRL)
- 8K context window (4x larger than text-only EmbeddingGemma)
- A commercially permissive Apache 2.0 license

[fatih c. akyon @fcakyon]

@VZHydra Eversince I started using OpenShip, my deployment cycles got faster, and my web app maintanence costs decreased more than 90 percent.

Now, an idea to fully scalable production deployment is max 2 days thanks to OpenShip 💯