Labops Meditechnology

Physicians juggle 5+ tools just to make one decision. Search evidence in one tab. Check drug interactions in another. Run the differential somewhere else. Code the visit in a fourth. Repeat, all day, every patient. Introducing Celiogen - an agentic AI workstation built to be the only tool open on a physician's desktop. Powered by Orinn-1.7, our state-of-the-art medical reasoning model, Celiogen reasons through evidence, differentials, drug decisions, and coding - as one connected system, not four disconnected ones. Built with a simple rule: your patient data is never retained. Not stored, not logged, not used to train anything. Ever. Launching early next week. We're opening registration for a limited early access, free credits, and a direct line to shape what we build next. $400 in credits for every physician who pre-registers early - link in the comments. #Celiogen #AgenticAI #PhysicianTools #ClinicalAI #HealthTech

We've just shipped a new feature in the Orinn API Platform Playground 🚀 - now in Beta. Starting today, you can: → Generate Bar, Line, and Pie charts directly in your Playground chat → Keep your data, analysis, and visuals all in one place → Export full conversations - charts included - as PDF or DOCX No extra tools. No copy-paste. No context switching. This is Beta, so we're actively improving it based on what we hear from users. If you try it, we'd genuinely love your feedback - what's working, what's missing, what we should build next. #Orinn #Labops #ProductUpdate

Anthropic just dropped Fable 5 - and yes, it beats our Orinn-1.7 on HealthBench Pro. 66 vs 64.8 -- Fair !! But here's what the benchmark slide doesn't show: Anthropic Fable 5 → $50/1M output tokens Labops Meditechnology Orinn-1.7 → $16/1M output tokens That's a 3x cost difference for a 1.8-point edge. For health AI at scale, that math matters - a lot.

We made Orinn 3× faster. 🚀 When you're building a super medical intelligence, two things are non-negotiable: data privacy and accuracy. We never compromise on either. But our customers were telling us Orinn was slow. And honestly? We knew it too. So we got to work. Today, we're proud to announce Orinn-1.7 - the same model you trust, now ~3× faster, with zero compromise on accuracy. Here's what the benchmarks look like:  1K token input   → 183 TPS  10K token input  → 143 TPS  100K token input → 92 TPS Orinn-1.7 now stands among the fastest AI models in the market - benchmarked head-to-head against GPT, Claude, and Gemini. And we're not stopping there: We've also expanded our context window to 260K tokens, with a 1M context window actively in development. Fast. Private. Accurate. Built for medical workflow, Orinn-1.7 is here.  To every customer who gave us that honest feedback - thank you. This one's for you. #Labops #Orinn #MedicalAI #HealthcareInnovation #AIPerformance #DigitalHealth

Orinn 1.7 is live. The best medical coding and clinical workflow AI model. Now available on the Orinn API platform. Here's what changed: After Orinn 1.6, we went back to the drawing board. Not to iterate. To leap. Orinn 1.7 is purpose-built for two things and two things mostly: → Best-in-class medical coding accuracy → End-to-end medical workflow automation No generic AI trying to do healthcare on the side. This is healthcare-first, built from the ground up. Try it today - and get up to $200 in free API credits to test it yourself. No waitlist. No demo call. Just go. Benchmarks and API platform link in the comments 👇 #Orinn #API #AI #MedicalAI #Labops #LabopsMeditechnology

40 parallel clinical tool calls, One prompt. One shot Results: → Google Gemini 3.1 Pro — 93% → Labops Meditechnology Orinn-1.6 — 97% Today we benchmarked one of the top SOTA models on the market (Gemini 3.1 Pro) against our latest model, Orinn-1.6 — on tool-calling ability for clinical workflow. Why does this matter? In the age of agentic AI, tool calling isn't a nice-to-have — it's the backbone. And in sensitive domains like healthcare, getting the wrong tool call doesn't just cause a bad UX. It can cause harm. The model needs to pick the right function, with the right arguments, in the right order — every single time. So we didn't make this easy. We built a 7-suite hard-mode benchmark designed to break models, not flatter them: 𝗦𝘂𝗶𝘁𝗲 𝟭 — 𝗜𝗺𝗽𝗹𝗶𝗰𝗶𝘁 𝗧𝗼𝗼𝗹 𝗖𝗮𝗹𝗹𝗶𝗻𝗴 No tool names in the prompt. "Assess her metabolic risk" — the model has to reason which tools to call on its own. 𝗦𝘂𝗶𝘁𝗲 𝟮 — 𝗠𝗲𝘀𝘀𝘆 𝗡𝗮𝗿𝗿𝗮𝘁𝗶𝘃𝗲 𝗘𝘅𝘁𝗿𝗮𝗰𝘁𝗶𝗼𝗻 Real doctor-speak: "she's about 5'4, weight's crept up to 82 kilos, her A1c came back at 7.8." Extract precise args from messy clinical notes. 𝗦𝘂𝗶𝘁𝗲 𝟯 — 𝗦𝗮𝗳𝗲𝘁𝘆-𝗙𝗶𝗿𝘀𝘁 𝗢𝗿𝗱𝗲𝗿𝗶𝗻𝗴 Patient has a penicillin allergy → doctor asks for amoxicillin. Does the model check the allergy BEFORE writing the prescription? Order matters. 𝗦𝘂𝗶𝘁𝗲 𝟰 — 𝗧𝗼𝗼𝗹 𝗗𝗲𝗽𝗲𝗻𝗱𝗲𝗻𝗰𝘆 𝗖𝗵𝗮𝗶𝗻𝘀 Tool B needs Tool A's output. "Calculate kidney function, then use that to adjust the vancomycin dose." The model must respect the chain. 𝗦𝘂𝗶𝘁𝗲 𝟱 — 𝗗𝗶𝘀𝘁𝗿𝗮𝗰𝘁𝗼𝗿 𝗥𝗲𝘀𝗶𝘀𝘁𝗮𝗻𝗰𝗲 20 tools available, only 1 needed. "Just calculate BMI." Does the model resist over-calling? Penalty for unnecessary tools. 𝗦𝘂𝗶𝘁𝗲 𝟲 — 𝗥𝗲𝗽𝗲𝗮𝘁𝗲𝗱 𝗧𝗼𝗼𝗹 𝗖𝗮𝗹𝗹𝘀 "Check warfarin interactions with ALL 4 of his other meds." That's 4 calls to the same function with different args. Most models collapse this into one. 𝗦𝘂𝗶𝘁𝗲 𝟳 — 𝗣𝗮𝗿𝗮𝗺𝗲𝘁𝗲𝗿 𝗣𝗿𝗲𝗰𝗶𝘀𝗶𝗼𝗻 "Patient is 172cm tall and weighs 68kg." Did the model swap height and weight? "Follow-up in 6 weeks" — did it convert to 42 days? Scoring: 35% tool selection + 35% argument precision + 15% over-calling penalty + 15% safety ordering. 20 tests. 7 suites. No shortcuts. Both models performed exceptionally well - which honestly says a lot about where the industry is heading. But in healthcare, the difference between 93% and 97% isn't a rounding error. It's the margin between a safe workflow and a missed allergy check. If you're building agentic systems in healthcare, try it against your model. The easy benchmarks are over. Time to test what actually matters. #AI #LLM #ToolCalling #MedicalAI #Orinn #Gemini #Labops Special thanks to Javier Rosas for evaluating the capability of our model earlier.  Benchmark code (open-source) → Check Comments 👇 Debditya Mallick Roslyn Singh Ahluwalia

The Orinn API is now live !!!! https://platform.orinn.ai/ Built for developers who are building the future of Medical AI. Orinn-1.6 topped every major medical benchmark - and now you can plug that intelligence directly into your product. What you get: 📨 Single API endpoint - simple, safer and reliable integration 🧠 Medical reasoning that outperforms GPT-5.2, Gemini-3-Pro, and more 📋 Clinical document understanding - EHRs, prescriptions, lab reports 👁️ Medical vision - interpret medical images, scans, and visual data 🚫 Lowest hallucination rate among all tested models Whether you're building clinical decision support, medical research tools, diagnostic platforms, or patient-facing AI - Orinn was built for exactly this. Docs are live. Go build something that matters. #Orinn #API #AI #MedicalAI #Labops #LabopsMeditechnology

Today we’re announcing Orinn-1.6 - our latest model built for medical and life sciences. 🧬 Orinn-1.6 is now the #1 AI model for Medical and Life Sciences. Here's what it looks like: 🏥 HealthBench → 71.2 (outperforming GPT-5.2-High, Gemini-3-Pro, Baichuan-M3) 🧠 MedQA → 95.8 📄 MedMCQA → 96.5 📊 PubMedQA → 95.3 🔬 MMLU Medical → 92.5 👁️ SLAKE (Medical VQA) → 85.5 📑 OmniDocBench → 88.5 Lowest hallucination rate among all tested models. That matters when lives are on the line. We didn't build Orinn to win benchmarks. We built it because healthcare deserves AI that doesn't guess - it knows. If you're building in: •⁠ ⁠Health-tech •⁠ ⁠Clinical software •⁠ ⁠Medical research tools •⁠ ⁠EHR / EMR platforms We'd love to work with you. Become an early partner. API platform launching soon. 🔥 #Orinn #LabopsMeditechnology #Healthtech #MedicalAI #AIforEHR

Labops Meditechnology is a member of NVIDIA Inception Program This milestone marks an exciting step in our journey to build the medical-grade alternative to generic LLMs. As part of #NVIDIAInception, we’ll utilize NVIDIA’s powerful GPU (H200) infrastructure to train our next SOTA model ORINN 2.0 - further enhancing its deep reasoning, understanding, and personalization capabilities. A warm thanks to the NVIDIA team for recognizing our mission and welcoming us to this global community of innovators driving the future of AI. Here’s to building intelligence that truly cares. #NVIDIAInception #AI #HealthcareAI #Labops #ORINN #Innovation