Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Moondream 2B running on an iPhone

vik

20,935 subscribers

33,412 views • 1 year ago •via X (Twitter)

Education Science & Technology

Anya Rossi• Live Now

Private livecam show

11 Comments

vik1 year ago

will be releasing ported weights + inference code soon, stay tuned!

baki1 year ago

why is there a Minecraft video on right lol, are we on TikTok?

vik1 year ago

my apologies, will do subway surfer next time

Charan1 year ago

I thought moondream was some model which generates game regions. I was waiting for it to pop on the Minecraft UI.

vik1 year ago

sorry... just wanted to give people something to look at while the spinner was spinning

ΜΛΛNΙ1 year ago

wat?! this is almost temporally perfect how did you do it before deepmind?! what's your food and sleep routine?! and how can I get this good?!

vik1 year ago

they have a model running on mobile but it sucks food/sleep -- i don't do either

Andrew Ngmi1 year ago

Is there a MLX port?

vik1 year ago

using coreml here

Maziyar PANAHI1 year ago

Pretty cool! Would love to learn how to fine-tune this for an on-device ocr!

vik1 year ago

working on an OCR-focused finetune so hopefully you won't have to do it yourself! what languages are you interested in?

Related Videos

Twilight Princess running native on an iPhone 17 at 120 fps

Twilight Princess running native on an iPhone 17 at 120 fps

Pheenoh

342,598 views • 1 month ago

The new Qwen 3.5 by Qwen running on-device on iPhone 17 Pro. Qwen 3.5 beats models 4 times its size, has strong visual understanding, and can toggle reasoning on or off. The 2B 6-bit model here is running with MLX optimized for Apple Silicon.

The new Qwen 3.5 by Qwen running on-device on iPhone 17 Pro. Qwen 3.5 beats models 4 times its size, has strong visual understanding, and can toggle reasoning on or off. The 2B 6-bit model here is running with MLX optimized for Apple Silicon.

Adrien Grondin

3,551,792 views • 3 months ago

Quantized Gemma 2B runs pretty fast on my iPhone 15 pro in MLX Swift. code & docs: Comparable to GPT 3.5 turbo and Mixtral 8x7B in LMSYS Org benchmarks but runs efficiently on an iPhone. Pretty wild.

Quantized Gemma 2B runs pretty fast on my iPhone 15 pro in MLX Swift. code & docs: Comparable to GPT 3.5 turbo and Mixtral 8x7B in LMSYS Org benchmarks but runs efficiently on an iPhone. Pretty wild.

Awni Hannun

79,702 views • 1 year ago

Releasing moondream-zig! It is a fast, implementation of moondream2 inference on the CPU written from-scratch in Zig :) moondream-zig provides 1.5-2x faster inference compared to huggingface on the same device. moondream vik

Releasing moondream-zig! It is a fast, implementation of moondream2 inference on the CPU written from-scratch in Zig :) moondream-zig provides 1.5-2x faster inference compared to huggingface on the same device. moondream vik

snow

29,245 views • 1 year ago

New Moondream 2B release! ✨ New features: - Long-form captioning - Open vocab tagging - Better counting, object detection, text understanding - Faster HF transformers inference

New Moondream 2B release! ✨ New features: - Long-form captioning - Open vocab tagging - Better counting, object detection, text understanding - Faster HF transformers inference

vik

51,735 views • 1 year ago

Running Qwen3 8B thinking on an iPhone Air with MLX. The model is quantized to 4-bit and runs pretty well.

Running Qwen3 8B thinking on an iPhone Air with MLX. The model is quantized to 4-bit and runs pretty well.

Awni Hannun

215,529 views • 8 months ago

Device Speed & Lag Example of the lag seen on an IPhone 15 PM running iOS 26 when configuring Lock Screen #iOS26

Device Speed & Lag Example of the lag seen on an IPhone 15 PM running iOS 26 when configuring Lock Screen #iOS26

Is iOS Stable?

13,460 views • 11 months ago

Excited to release a preview of Moondream 3. A 9B param, 2B active MoE vision language model that makes no compromises; offering state-of-the-art visual reasoning while still retaining an efficient and deployment-friendly form factor.

Excited to release a preview of Moondream 3. A 9B param, 2B active MoE vision language model that makes no compromises; offering state-of-the-art visual reasoning while still retaining an efficient and deployment-friendly form factor.

vik

227,375 views • 8 months ago

Qwen3.5-2B vs Qwen3.5-9B rap battle running locally on 2 foidable phones

Qwen3.5-2B vs Qwen3.5-9B rap battle running locally on 2 foidable phones

alexine 🏴‍☠️

134,212 views • 3 months ago

Alibaba’s Qwen 3.5 is now running fully on-device on the iPhone 17 Pro. It outperforms models 4x its size, delivers strong visual understanding, and lets you switch reasoning on or off. This demo uses the 2B 6-bit version, optimized with MLX for Apple Silicon.

Alibaba’s Qwen 3.5 is now running fully on-device on the iPhone 17 Pro. It outperforms models 4x its size, delivers strong visual understanding, and lets you switch reasoning on or off. This demo uses the 2B 6-bit version, optimized with MLX for Apple Silicon.

Hugging Models

202,648 views • 2 months ago

"It's an iPhone -4!" 🤣 Roy upgrading his phone might be football’s longest-running saga!📱

"It's an iPhone -4!" 🤣 Roy upgrading his phone might be football’s longest-running saga!📱

The Overlap

232,357 views • 14 days ago

⚡️New "Thinking" UI with DeepSeek R1 Distilled models, running completely locally on iPhone

⚡️New "Thinking" UI with DeepSeek R1 Distilled models, running completely locally on iPhone

Private LLM

188,633 views • 1 year ago

At 0x41con, I talked about how a newer version of Trigon was developed for A7 - A9 and A11 devices. I demoed the following untethered iOS 14 jailbreak, which is 100% reliable on all devices thanks to Trigon - the following video shows it running on an iPhone 6S and an iPhone 7.

At 0x41con, I talked about how a newer version of Trigon was developed for A7 - A9 and A11 devices. I demoed the following untethered iOS 14 jailbreak, which is 100% reliable on all devices thanks to Trigon - the following video shows it running on an iPhone 6S and an iPhone 7.

Alfie

58,606 views • 1 year ago

Running 400B model on iPhone! 0.6 t/s Credit Dan Woods @alexintosh Daniel Isaac Anemll

Running 400B model on iPhone! 0.6 t/s Credit Dan Woods @alexintosh Daniel Isaac Anemll

Anemll

291,054 views • 2 months ago

This is using an iPhone 14 basic, but it can be used on an iPhone 12-15 sc ig/angel_sensia

This is using an iPhone 14 basic, but it can be used on an iPhone 12-15 sc ig/angel_sensia

Mutia

139,653 views • 7 months ago

iPhone running at 1 Hz be like:

iPhone running at 1 Hz be like:

Noah Cat

194,329 views • 1 year ago

🇵🇸🇮🇱 Doctor’s in Gaza are having to operate under the flashlight of an iPhone due to power running low.

🇵🇸🇮🇱 Doctor’s in Gaza are having to operate under the flashlight of an iPhone due to power running low.

Censored Men

5,628,372 views • 2 years ago

Updated my HF Space for vibe testing smol VLMs on object detection, visual grounding, keypoint detection & counting! 👓 🆕Compare Qwen2.5 VL 3B vs Moondream 2B side-by-side with annotated images & text outputs. Try examples or test your own images! 🏃👇

Updated my HF Space for vibe testing smol VLMs on object detection, visual grounding, keypoint detection & counting! 👓 🆕Compare Qwen2.5 VL 3B vs Moondream 2B side-by-side with annotated images & text outputs. Try examples or test your own images! 🏃👇

Sergio Paniego

15,717 views • 11 months ago

Ronaldo’s thoughts on buying his son an iPhone.

Ronaldo’s thoughts on buying his son an iPhone.

Deepak Prabhu

480,220 views • 1 year ago

Announcing moondream 0.5B, the world's smallest vision language model.

Announcing moondream 0.5B, the world's smallest vision language model.

vik

58,888 views • 1 year ago