376 Photos and videos
No data, no problem
introducing agentic synthetic data generation with Cosmos 3
share a few examples, generate more data, automate model training, automatically deploy the latest version with no downtime
in a benchmark run with Corning Incorporated's optical fiber manufacturing engineering team, a model trained on 8 real defect images plus synthetic examples generated by Cosmos reached 0.95 mean average precision and perfect recall on the toughest defect class, beating a baseline trained on real data alone.
"The Roboflow Agent powered by NVIDIA allows us to generate the training data we need, fine-tune our models, and strengthen model performance and inspection quality while increasing the speed, scalability, and adoption of next-generation technologies,” - Jeremy Knopf, chief information officer, Corning Optical Communications
Introducing Cosmos 3: Our latest frontier model for Physical AI
Cosmos 3 is the world’s first fully open omnimodel with native vision reasoning, world and action generation.
Today we’re releasing Super (32B) and Nano (8B) variants.
7
29
174
38,067
Roboflow retweeted
Jun 11
looks like we are working on the same thing haha; more soon
Jun 11
We’re teaming up @Palmeiras, the first football club to meaningfully build upon TacticAI: our AI system that can help simulate field scenarios and predict open play dynamics up to 8 seconds in advance. ⚽
55
178
3,610
647,152
You can now build vision applications with Fable 5 in Roboflow
Fable 5 is the best Anthropic model for visual understanding, beating Opus 4.8 on our evals
very high object understanding capability means it's a great option when using a VLM as a judge in your pipeline
ideal for a final step in a process that requires judgement or interpretation of a situation
of course, nails most document, label extraction, ocr, scene reports, etc that you expect of a large frontier model
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
Its capabilities exceed those of any model we’ve ever made generally available.
3
2
43
6,074
breakdown of Fable 5 eval results and findings
blog.roboflow.com/claude-fab…
1
954
try Fable 5 against other frontier models on vision tasks
playground.roboflow.com/obje…
189
supervision is #3 trending repo today and crossed 40k stars a few days ago, we built it to make computer vision dead simple
what would make it even easier? what is slowing you down when building with vision tools?
let us know the features you want, we are working on the roadmap. need your feedback
Jun 6
supervision just hit 40,000 GitHub stars!
it now powers over 6.5k open-source computer vision projects, including all my demos like basketball AI
link: github.com/roboflow/supervis…
4
37
2,975
No data, no problem
introducing agentic synthetic data generation with Cosmos 3
share a few examples, generate more data, automate model training, automatically deploy the latest version with no downtime
in a benchmark run with Corning Incorporated's optical fiber manufacturing engineering team, a model trained on 8 real defect images plus synthetic examples generated by Cosmos reached 0.95 mean average precision and perfect recall on the toughest defect class, beating a baseline trained on real data alone.
"The Roboflow Agent powered by NVIDIA allows us to generate the training data we need, fine-tune our models, and strengthen model performance and inspection quality while increasing the speed, scalability, and adoption of next-generation technologies,” - Jeremy Knopf, chief information officer, Corning Optical Communications
Introducing Cosmos 3: Our latest frontier model for Physical AI
Cosmos 3 is the world’s first fully open omnimodel with native vision reasoning, world and action generation.
Today we’re releasing Super (32B) and Nano (8B) variants.
7
29
174
38,067
and cosmos 3
nvidianews.nvidia.com/news/n…
1
575
good thread on using cosmos 3 for visual understanding and reasoning
x.com/erik_kokalj/status/206…
Jun 1
Tested NVIDIA's new Cosmos 3 myself. Easy tier first: zero-shot gate turnaround.
Single inference of 1.5min video, 0.5fps ingestion, 6 segmented states. Thinking on, no fine-tuning, prompted it to focus on cargo in the cropped ROI.
Harder tests below. 🧵
719
rf-detr helping make big gains for small object tracking
detection rate from 47% to 86%
Swapped my ball tracker from TrackNet (3-frame temporal CNN) to RF-DETR (transformer, single-frame). Trained on ~4,700 labeled pickleball frames from public data. Detection rate jumped from 47% to 86%.
2
1
27
2,714
Roboflow retweeted
May 27
RF-DETR is now available in @huggingface transformers
state of the art in both detection and segmentation, outperforming YOLO architectures
- checkpoints: huggingface.co/Roboflow/mode…
- demo: huggingface.co/spaces/huggin…
- docs: huggingface.co/collections/m…
25
123
1,171
78,931
You can now use models on OpenRouter directly in Roboflow
New models are dropping weekly and it's important to build your pipeline around the problem you're solving. OpenRouter in Roboflow lets you do that.
Plug in whatever model is winning this month, and swap it out the week after when something better lands or find a cheaper variant that works well enough
May 26
Today we’re announcing our $113M Series B led by @CapitalGVC.
Over the last 6 months, weekly volume on OpenRouter grew from 5T to 25T tokens as AI rapidly shifts from experimentation into production.
We’re excited for what comes next.
2
2
16
2,633
we saw the same thing with our vision leaderboard
still testing and wrapping our heads around Gemini 3.5 Flash but it feels like a pivotal shift in frontier models
way cheaper and way faster than anything else, and performance is super high
first time it feels like you can really use a VLM for high scale tasks. lots of previous tasks that required making workflows of smaller, cheaper models now looks possible (speed and cost wise) with Gemini 3.5 Flash
a few examples below
May 21
Gemini 3.5 Flash ranks #1 on Automation Bench (from Zapier), beating every other frontier model at a much lower cost
5
12
144
17,094
stuff like this relative reference example enables for creating application logic with only text inputs
In the past you would need a custom pipeline to understand cars relative to specific lanes would have required creating zones specific to the lanes
Gemini 3.5 Flash can do this automatically
1
13
1,515