Anderson retweeted
Most AI agents say "done." Nobody checks if it's true.
SoduWork flips that. It's an open-source workbench where:
→ A human sets the acceptance criteria
→ Agents do the work
→ No evidence, no pass
→ A different model reviews it
Worth a look 👇
➪ github.com/Mdiri-AI-Public/S…
🤝 Paid partnership · 🤖 Made with AI
It's exciting to see medical AI models focusing on complete workflows.
LingSante-3.1-air-VL can analyze images, generate reports, find related research, and navigate healthcare apps.
The combination of specialized knowledge and tool use opens up interesting possibilities.
Meet LingSante-3.1-air-VL, an open-source multimodal medical model with 124B total and 5.5B active parameters. It combines medical image understanding, report generation and medical reasoning with multimodal search and tool use.
In our evaluation, it leads the compared models on AFUMED-Derm, AFUMED-Report, MIMIC-CXR and medical VQA (averaged across benchmarks), ahead of GPT-6.1 Sol, Gemini 3.8 Flash on all four.
It also demonstrates strong general-purpose capabilities, scoring higher than GPT-6.1 Sol on BrowseComp-VL and Gemini 3.8 Flash on AndroidWorld, alongside competitive performance on MMSearch. Explore LingSante-3.1-air-VL and build with it.
Medical AI is moving fast, and open-weight models are becoming increasingly capable.
LingSante-3.1-flash combines 560B total parameters, advanced medical reasoning, and professional workflow execution.
This is a strong step toward more capable healthcare AI.
Today, we release LingSante-3.1-flash, an open-weight MoE model with 560B total parameters and 25B activations per token. It unifies diverse medical capabilities while retaining general capabilities, including complex medical reasoning, long-horizon workflow execution, and professional behavior alignment.
LingSante-3.1-flash moves beyond medical QA toward real-world healthcare problem. The model leads on HealthBench Professional (65.72), HLE (text-only, with tools: 72.22) and BrowseComp (with context: 94.10), outperforming GPT-6 Astra, Claude Opus 5.5 and Gemini 4 Argon on all three. LingSante-3.1-flash also performs well on professional long-horizon medical workflows, such as PhysicianBench (57) and EHR-Complex (74.40).
Together, these results establish LingSante-3.1-flash as the leading open-weight models for healthcare.
The weights are available. Explore LingSante and build with it.
More below ↓
1
3,983
Anderson retweeted
Ever saved a hundred references but still had no idea how to turn them into something of your own? Same. 😭
I've been using CapCut Web to explore visual ideas and turn inspiration into images I can actually work with.
➪ capcut.com/my-edit?utm_mediu…
#CapCutWeb #AIImage
🤝 Paid partnership · 🤖 Made with AI
Ever saved a hundred references but still had no idea how to turn them into something of your own? Same. 😭
I've been using CapCut Web to explore visual ideas and turn inspiration into images I can actually work with.
➪ capcut.com/my-edit?utm_mediu…
#CapCutWeb #AIImage
🤝 Paid partnership · 🤖 Made with AI
Instead of getting stuck trying to recreate one reference perfectly, I can experiment with different layouts, moods, and styles.
It's a fun way to discover a visual direction I might not have thought of before.
🤝 Paid partnership
1
3,041
A little creative inspiration goes a long way.
From Nov 20–26, new or upgrading CapCut Web Ultra users get 7 days of unlimited image generation on select models. Time to bring those saved ideas to life!
#CapCutWebDesignStudio #CapCutWebVideoEditor
🤝 Paid partnership
2,115
Most AI agents say "done." Nobody checks if it's true.
SoduWork flips that. It's an open-source workbench where:
→ A human sets the acceptance criteria
→ Agents do the work
→ No evidence, no pass
→ A different model reviews it
Worth a look 👇
➪ github.com/Mdiri-AI-Public/S…
🤝 Paid partnership · 🤖 Made with AI
thought this would be startups selling to startups
nope. one in six buyers in the dataset had 1,000+ employees
The best sales teams win 5x as often.
We analyzed $1.2B+ in pipeline, 10,000+ deals, and 170,000+ emails and calls on @reevo_ai to find out how they do it.
Check it out below 👇
1
106
Anderson retweeted
my favourite detail in sciforium's writeup: they tried all-reduce on its own stream, it fought the other ops for memory bandwidth, and they said it didn't help
headline is DeepSeek-V4.1-Flash at 1,129 tok/s on four AMD GPUs
🤝 Paid partnership
Anderson retweeted
When NVIDIA, AMD, Intel, Google and Meta get behind the same open source AI project, you know something big is happening. 👀
SGLang Summit 2026 brought major players together to shape the future of AI inference
This is bigger than a summit. It's a major shift for open source AI
2,000+ people. Two days. The first SGLang Summit.
Together, we announced SGLang v1.0, released Miles v0.2, and launched the SGLang Consortium with nine founding members: @nvidia, @AMD, @intel, @Google, @Meta, @thinkymachines, @Cloudflare, @modal, and @radixark.
SGLang is now used by 332 organizations. At Meta, it powers production inference for products serving billions of users. Teams from Modal, Cloudflare, LinkedIn, and others shared their production lessons.
Across 90+ technical talks and six hands-on workshops, the community explored what comes next for inference, reinforcement learning, and AI agents.
To everyone building, deploying, and contributing: thank you for making this possible.
Explore the announcements, engineering lessons, and moments from SGLang Summit 2026 in the link below.
my favourite detail in sciforium's writeup: they tried all-reduce on its own stream, it fought the other ops for memory bandwidth, and they said it didn't help
headline is DeepSeek-V4.1-Flash at 1,129 tok/s on four AMD GPUs
🤝 Paid partnership
When NVIDIA, AMD, Intel, Google and Meta get behind the same open source AI project, you know something big is happening. 👀
SGLang Summit 2026 brought major players together to shape the future of AI inference
This is bigger than a summit. It's a major shift for open source AI
2,000+ people. Two days. The first SGLang Summit.
Together, we announced SGLang v1.0, released Miles v0.2, and launched the SGLang Consortium with nine founding members: @nvidia, @AMD, @intel, @Google, @Meta, @thinkymachines, @Cloudflare, @modal, and @radixark.
SGLang is now used by 332 organizations. At Meta, it powers production inference for products serving billions of users. Teams from Modal, Cloudflare, LinkedIn, and others shared their production lessons.
Across 90+ technical talks and six hands-on workshops, the community explored what comes next for inference, reinforcement learning, and AI agents.
To everyone building, deploying, and contributing: thank you for making this possible.
Explore the announcements, engineering lessons, and moments from SGLang Summit 2026 in the link below.
everyone who said location data was dead owes this team an apology
Introducing the CIA for Meta ads.
You no longer need to rely on Meta’s Algorithm.
For the first time, you can choose the exact devices your ads go to.
Name a store, event or any location and Punk shows your Meta ads to the phones that were there.
Built by talent from Anthropic, BCG and… a rave DJ.
First 100 users get a cracked campaign built 1:1 with me.
2,550
Launching your first AI agent isn’t the finish line.
When it fails, can your team find the issue, test a fix, and improve it without a vendor?
That’s what Retell is solving.
The future belongs to teams that keep making their AI better.
Retell just crossed $100M in revenue run rate.
$10M to $100M in 17 months. Profitable. $4.6M raised.
From day one, we’ve focused on solving the hardest parts of deploying voice AI in the product itself, so customers can build, manage, and improve their own agents without depending on us.
Our bet: the future of the contact center belongs to the teams who run it.
Proud of our team. Grateful to every customer who believed in us.
trying Giotto from the OpenAI SDK is two lines and an API key:
base_url = ""cloud.giotto.ai/v1""
model = ""giotto-solo""
Qoduck, a new stealth model, is now available in Qoder.
Qoder is Alibaba’s coding agent (from the team behind Qwen). You can download it and try Qoduck free for a limited time.
qoder.com
The real flex isn’t 560B parameters.
It’s turning PDFs into working code, financials into board decks, and 2D images into rich 3D scenes.
That’s why Ling-3.2-flash matters: less hype, more results.
Introducing Ling-3.2-flash, Ant Ling's native multimodal model: 560B total / 25B active parameters.
Stronger complex visual understanding, long-horizon multimodal agents, 3D understanding & interaction, and medical intelligence.
Outscores Opus 5 (max) in release evaluations:
- OmniDocBench1.5: 92.21 vs 90.56
- ZEROBench(Sub): 43.71 vs 39.52
- Vision2Web: 85.86 vs 80.20
🚨 A code reviewer that reproduces the crash.
On a copy of Sentry's code, Aether's reviewer hit the crash itself and posted the actual error with its finding. Not a guess from reading the code.
🤝 Paid partnership
Today we’re introducing Aether @aetherruntime.
The fastest cloud coding agents in the world that you can use with your existing subscription
Give it a task and just 20 seconds later your build is running on its own cloud machine. You can pick up from your phone, watch every line, and step in whenever you want as if it was local. Close the laptop. It keeps going.
It runs Claude Code or Codex on the Claude or ChatGPT plan you already pay for so there’s not a 2nd AI bill. You only pay for the machine.
When it’s done, if code review is enabled, a reviewer agent runs the code with the changes, which catches classes of bugs that can't be found from static analysis.
Before building Aether I built infra at Apple, Uber, and HeyGen. I built Aether because the cloud agents I tried before all had drawbacks, and I couldn't imagine building something without it now.
It’s free to start. runaether.dev
2
111
most restyling tools I've used start drifting halfway through, the face turns into someone else and the props move around.
Viggle Remake claims the acting and camera stay locked for the whole video, so I'd throw a long product walkthrough at it first
🤝 Paid partnership
Introducing Viggle Remake.
The most powerful video transformation agent. And we’re releasing the weights today.
→ recast every character in a scene, and they stay consistent across cuts
→ swap objects, change the background, restyle or add VFX
→ up to 2 minutes
→ keep chatting to fix any shot until it’s right
From $0.96 per 10 seconds at 720p, less than half of what Higgsfield Genjutsu charges.
Try it now: viggle.ai/meme/app/home?ui=v…
Open weights: huggingface.co/Viggle/Viggle…



















