Shaping the future of AI development
Compute, community, and cutting-edge research for AI developers defining what's next
Building the future of AI through research and mentorship
Neural Software: from vision to reality
Join Lambda's co-founder and CEO Stephen Balaban as he unveils Neural Software: an entirely new way of thinking about software that can collaborate with humans, evolve over time, and adapt like never before.
Our work at NeurIPS 2025
Scaling laws for diffusion models
Diffusion beats autoregressive in data-constrained settings.
SpatialReasoner
Builds an explicit 3D scene and reasons over it step by step, boosting accuracy and generalization on 3D spatial QA benchmarks.
Tensor decomposition for force-field prediction
Replaces heavy tensor operations in molecular force-field models with low-rank approximation, reducing compute while keeping accuracy.
Bifrost-1
Aligns VLMs with diffusion models through shared CLIP patch embeddings, enabling controllable high-quality generation while preserving reasoning.
BLEUBERI
Using simple BLEU scores as feedback on hard instructions can train instruction-following models that rival those tuned with expensive learned rewards.
OverLayBench
A dataset that stress-tests layout-to-image models on heavily overlapping scenes, exposing current failures and offering an improved baseline.
The ARChitects secure runner-up in ARC Prize 2025
Last year, “the ARChitects,” a Lambda-sponsored team (including Lambda researcher David Hartmann) won the ARC Prize 2024. This year, they finished second out of 1,400+ teams with a final leaderboard score of 16.53%.
LLM performance benchmarks leaderboard
A clear, data-driven comparison of today's leading large language models. Standardized benchmark results cover top contenders like Meta's Llama 4 series, Alibaba's Qwen3, and the latest from DeepSeek, with critical performance metrics measuring everything from coding ability to general knowledge.
ML Times
Your go-to source for the latest in the field, curated by AI. Sift through the excess. Make every word count.
Best practices and system insights
Diffusion from scratch
A guide for diffusion models implemented in a single PyTorch script.
Text2Video pretraining
Lessons learned from training a text-to-video model with hundreds of GPUs.
GPU benchmarks
Throughput GPU benchmarks for training deep neural networks.
MLCommon benchmark
Time-to-solution benchmark for training foundation models on clusters.
Recognized by scholars and industry peers
ICML 2025 Latent thought models
Teach LLMs to “think” before they speak and improve parameter and memory efficiency. ICML 2025 Product of experts with LLMs
Boosting performance on the ARC-AGI challenge is a matter of perspective.
* Winner, ARC-AGI 2024 ICCV 2025 DepR
Uses depth cues and diffusion models to turn a single image into a clean, instance-level 3D scene. ICCV 2025 Video MMLU
A benchmark that tests whether models can truly follow and understand long, multi-subject lecture videos, revealing that current VLM’s limitations. CoRL 2025 Latent adaptive planner
Robots can learn a compact “plan space” from human demos and continually updates its plan as the scene changes, enabling faster and more reliable manipulation. CoRL 2025 AimBot
AimBot draws visual cue onto robot camera images so policies directly see where the gripper is in 3D, boosting manipulation success. EMNLP 2025 Word salad chopper
Detects when a reasoning model has drifted into meaningless repetition and cleanly cuts away those extra tokens, saving a lot of output cost with almost no quality loss. EMNLP 2025 DEL-ToM for theory-of-mind reasoning
Breaks social reasoning into “who-knows-what” steps and uses a checker to pick the most consistent answer, improving small-models’ theory-of-mind performance. EMNLP 2025 VeriFastScore
Trains a single model to extract and verify all claims in a long answer at once using web evidence, speeding up long-form factuality evaluation. EMNLP 2025 Error typing for smarter rewards
Labeling each reasoning step’s mistake type and converts those labels into scores, giving richer feedback and improving math problem solving with less data.
Breakthroughs backed by Lambda
Bold ideas, funded and refined through the Lambda Research Grant. These are the projects shaping how AI learns, reasons, and scales — built by the researchers defining what’s next.
SAEBench
A comprehensive benchmark for sparse autoencoders in language model interpretability
Adam Karvonen, Can Rager, Johnny Lin, Curt Tigges, Joseph Bloom, David Chanin, Yeu-Tong Lau, Eoin Farrell, Callum McDougall, Kola Ayonrinde, Demian Till, Matthew Wearden, Arthur Conmy, Samuel Marks, and Neel Nanda — ICML 2025
VideoHallu
Evaluating and mitigating multi-modal hallucinations on synthetic video understanding
Zongxia Li, Xiyang Wu, Guangyao Shi, Yubin Qin, Hongyang Du, Tianyi Zhou, Dinesh Manocha, and Jordan Lee Boyd-Graber — NeurIPS 2025
VLM2Vec-V2
Advancing multimodal embedding for videos, images, and visual documents
Meng, Rui and Jiang, Ziyan and Liu, Ye and Su, Mingyi and Yang, Xinyi and Fu, Yuepeng and Qin, Can and Chen, Zeyuan and Xu, Ran and Xiong, Caiming, and others — arXiv preprint 2025
Think, prune, train, improve
Scaling reasoning without scaling models
Caia Costello, Simon Guo, Anna Goldie, and Azalia Mirhoseini — ICLR 2025 workshop
NeoBERT
A next-generation BERT
Lola Le Breton, Quentin Fournier, Mariam El Mezouar, and Sarath Chandar — TMLR 2025
Join us in shaping the future of AI
We're committed to supporting groundbreaking research by offering qualifying researchers up to $5,000 in cloud credits to develop and showcase their work using Lambda's Instances, with select research to be featured on our website.