Published AI and science developmentsAI & Science

Neuro-VL is a vision-language model trained on nearly one million routine clinical neuroimaging studies that can localize, describe, diagnose and report on complete examinations and discuss its findings in natural language. On NeuroBench, built from internal studies and 24 international public datasets, it outperformed GPT-5.4 and Gemma 4 on 16 of 17 task-group combinations and the neuroradiology-specialized NeuroVFM on all 17, leading by 15.9, 16.9 and 23.0 percentage points when averaged equally across combinations. A further-tuned longitudinal version compares prior and current studies, and the model resisted misleading diagnostic or localization prompts more reliably than GPT-5.4.

Neuro-VL is a vision-language model trained on nearly one million routine clinical neuroimaging studies that can localize, describe, diagnose and report on complete examinations and discuss its findings in natural language. On NeuroBench, built from internal studies and 24 international public datasets, it outperformed GPT-5.4 and Gemma 4 on 16 of 17 task-group combinations and the neuroradiology-specialized NeuroVFM on all 17, leading by 15.9, 16.9 and 23.0 percentage points when averaged equally across combinations. A further-tuned longitudinal version compares prior and current studies, and the model resisted misleading diagnostic or localization prompts more reliably than GPT-5.4.

P2Echo folds multi-view echocardiography segmentation into a single network that takes a structured text prompt naming the view and target structure and uses it to modulate decoder features. Previously each standard acoustic window typically needed its own specialist model, or relied on implicit priors that were hard to control explicitly. On internal data, P2Echo improves over EchoONE, the other multi-view model, by 5.3, 1.1 and 2.8 Dice on 2CH, 3CH and 4CH, is comparable on PSAX, and holds competitive results on three external datasets at 6.90 GFLOPs.

P2Echo folds multi-view echocardiography segmentation into a single network that takes a structured text prompt naming the view and target structure and uses it to modulate decoder features. Previously each standard acoustic window typically needed its own specialist model, or relied on implicit priors that were hard to control explicitly. On internal data, P2Echo improves over EchoONE, the other multi-view model, by 5.3, 1.1 and 2.8 Dice on 2CH, 3CH and 4CH, is comparable on PSAX, and holds competitive results on three external datasets at 6.90 GFLOPs.

Mixture-of-experts models must hold every expert in memory at inference even though each token uses only a few, so deleting the least important experts is a common compression route. HOPE writes the pruning error as a sum of pairwise expert interaction terms, builds an interaction matrix from calibration data, and solves a quadratic program to choose the prune-set. Across Qwen3.5-122B, Qwen3.5-35B and GLM-4.5-Air, HOPE holds an average rank of 1.58 at 50% pruning versus 2.42 for the next-best method, REAP; Qwen3.5-35B-A3B scores 42.3 on SWE-Bench Pro, matching the unpruned model's 41.8 while REAP falls to 36.2.

Mixture-of-experts models must hold every expert in memory at inference even though each token uses only a few, so deleting the least important experts is a common compression route. HOPE writes the pruning error as a sum of pairwise expert interaction terms, builds an interaction matrix from calibration data, and solves a quadratic program to choose the prune-set. Across Qwen3.5-122B, Qwen3.5-35B and GLM-4.5-Air, HOPE holds an average rank of 1.58 at 50% pruning versus 2.42 for the next-best method, REAP; Qwen3.5-35B-A3B scores 42.3 on SWE-Bench Pro, matching the unpruned model's 41.8 while REAP falls to 36.2.

IBM researchers released LLMON (LLM Object Notation), a markup language that uses reserved special tokens to mark the boundaries of instructions, data and requirements, together with two experimental preview models trained from Granite 4.1 3B and 8B base checkpoints. The approach turns each semantically coherent component of the context into an addressable block, and an execution block states which instruction to run, which data it operates on, and where the output goes. On the OPI benchmark the LLMON previews avoided 47% more attacks than the best frontier model tested and 95% more than the best open-source model, while general benchmark performance stayed on par with Granite 4.1 Instruct.

IBM researchers released LLMON (LLM Object Notation), a markup language that uses reserved special tokens to mark the boundaries of instructions, data and requirements, together with two experimental preview models trained from Granite 4.1 3B and 8B base checkpoints. The approach turns each semantically coherent component of the context into an addressable block, and an execution block states which instruction to run, which data it operates on, and where the output goes. On the OPI benchmark the LLMON previews avoided 47% more attacks than the best frontier model tested and 95% more than the best open-source model, while general benchmark performance stayed on par with Granite 4.1 Instruct.

Tether AI Research released BotPsy, a 0.8B-parameter model built on Qwen3.5-0.8B for function calling and agentic workflows that runs on-device. Trained through supervised fine-tuning, multi-teacher on-policy distillation and two reinforcement-learning rounds, it lifts BFCL v4 from its base's 23.4 to 44.0 and τ-bench from 23.2 to 51.5, and beats Qwen3.5-2B-Think, a model 2.5 times its size, on five of six tool-use and agentic benchmarks, trailing only on Claw-Eval (46.5 vs 48.9). It still makes mistakes, targets low-to-medium difficulty tasks and is not intended for high-stakes use.

Tether AI Research released BotPsy, a 0.8B-parameter model built on Qwen3.5-0.8B for function calling and agentic workflows that runs on-device. Trained through supervised fine-tuning, multi-teacher on-policy distillation and two reinforcement-learning rounds, it lifts BFCL v4 from its base's 23.4 to 44.0 and τ-bench from 23.2 to 51.5, and beats Qwen3.5-2B-Think, a model 2.5 times its size, on five of six tool-use and agentic benchmarks, trailing only on Claw-Eval (46.5 vs 48.9). It still makes mistakes, targets low-to-medium difficulty tasks and is not intended for high-stakes use.

PixVid converts a pretrained latent video diffusion model into a frame-wise pixel-space generator: it keeps the diffusion transformer, replaces the VAE decoder with a convolutional detailer head, and finetunes from Wan2.1-14B. It reaches VBench totals of 85.09 at 960p and 84.78 at 480p, above Wan2.1's 84.05 and PixelUMM's 83.72, and keeps 98% of source motion on fast clips versus 83% for an identically finetuned latent model.

PixVid converts a pretrained latent video diffusion model into a frame-wise pixel-space generator: it keeps the diffusion transformer, replaces the VAE decoder with a convolutional detailer head, and finetunes from Wan2.1-14B. It reaches VBench totals of 85.09 at 960p and 84.78 at 480p, above Wan2.1's 84.05 and PixelUMM's 83.72, and keeps 98% of source motion on fast clips versus 83% for an identically finetuned latent model.