We’ve fixed a bug that was degrading image understanding in GPT-6 Sol and GPT-6 Luna. You should now see better results on visual tasks in the API and Codex, including computer use.
Heading to #ECCV2026 in Malmö, Sept 8-13, presenting 4 works at MedFMB & AI4M3D: RadMatch (open source report evaluation), Jolia, Curia-2 & Curia-MAE, a new addition to the Curia family.
Come find our team there 🚀
Congrats to the whole team for their hard work! @SovanKhlt, @CharlesCorbiere, Baptiste Callard, Léo Butsanets, Théo Danielou, Antoine Saporta, Amaury Prat, Leo Machado, @Phylliade
At AI4M3D (Wednesday Morning):
Curia-MAE: Multi-Modal Multi-Anatomy MAE Pre-Training for 3D Medical Image Segmentation (arxiv.org/abs/2608.05844)
and we also present Jolia and Curia !
Malmö this week for #ECCV2026. Presenting RadMatch at the MedFM-Bench workshop. Come say hi if you work on medical foundation model or radiology report generation 👋
We'll be at @eccvconf 2026 in Malmö, September 8 to 13, presenting our latest ML research for radiology at the MedFMB and AI4M3D workshops.
More details on what we're sharing to come.
If you're attending, come find our team!
When working on radiology report generation, we noticed a gap in evaluation : a metric that was reliable and auditable, and could guide us to improve models. To fill this gap, we just released RadMatch !
Congrats to @CharlesCorbiere and the team, more to come soon!
Today we're sharing RadMatch, an open source framework that evaluates AI-generated radiology reports finding by finding, not just word overlap.
Accepted at the MedFMB workshop, ECCV 2026.
Full blogpost: raidium.eu/blog/RadMatch/?mt…#MedicalImaging#MLResearch
Today we're sharing RadMatch, an open source framework that evaluates AI-generated radiology reports finding by finding, not just word overlap.
Accepted at the MedFMB workshop, ECCV 2026.
Full blogpost: raidium.eu/blog/RadMatch/?mt…#MedicalImaging#MLResearch
We're heading to MICCAI 2026 in Strasbourg (Sept 27–Oct 1)!
Our team will share our latest foundation model research: an oral presentation of Curia-2, a keynote by @cdancette at the FOMO26 workshop, & an open-source evening event.
Connect with our team on site to chat!
In Taipei for #MIDL2026 this week! Excited to present our oral paper RadImageNet-VQA on Wednesday.
If you're around and want to chat about radiology foundation models or what we're building at @raidium_med, feel free to ping me!
🚨 𝗠𝗲𝗱𝗶𝗰𝗮𝗹 𝗔𝗜 𝗥𝗲𝘀𝗲𝗮𝗿𝗰𝗵 𝗔𝗹𝗲𝗿𝘁! 🚨
How can we improve 3D CT analysis by aligning vision and language at a finer, concept-level granularity, moving beyond global representations?
@Raidium presents 𝗝𝗼𝗹𝗶𝗮 with its 𝗖𝗼𝗻𝗤𝘂𝗲𝗿 𝗳𝗿𝗮𝗺𝗲𝘄𝗼𝗿𝗸: a 3D CT foundation model leveraging concept-level vision-language alignment without spatial supervision.
By Julien Khlaut, @CharlesCorbiere , Baptiste Callard, Amaury Prat, Leo Butsanets, Antoine Saporta, Théo Danielou, Leo Machado, Korentin Le Floch, Tom Boeken, Pierre Manceron and @cdancette
Now you can watch and listen to the latest Medical AI papers daily on our YouTube and Spotify channels!
YouTube Explainer: youtu.be/PgzM1XUeRhI
YouTube Shorts: youtube.com/shorts/f7O4ExLlZ…
Spotify: open.spotify.com/show/4edRuS…
Here's why it's exciting: 👇🧵 1/9
#MedicalAI#Healthcare#CTImaging#VisionLanguage
[1/9]