[metadata]
description: Unlock the power of compute with Lambda — the AI computing platform empowering developers to innovate, train models, and shape the future of superintelligence.
generator: HubSpot
google-site-verification: Yl9AOUwQUNp73F6qbOEGV8Z-ottfBhEf0Wn0VLsuYKE
og:description: Unlock the power of compute with Lambda — the AI computing platform empowering developers to innovate, train models, and shape the future of superintelligence.
og:image: https://lambda.ai/hubfs/OPENGRAPH%20-%20Research.png
og:image:height: 630
og:image:width: 1200
og:title: AI Research & Publications on AI Computing Platform | Lambda
og:url: https://lambda.ai/research
twitter:card: summary_large_image
twitter:description: Unlock the power of compute with Lambda — the AI computing platform empowering developers to innovate, train models, and shape the future of superintelligence.
twitter:image: https://lambda.ai/hubfs/OPENGRAPH%20-%20Research.png
twitter:title: AI Research & Publications on AI Computing Platform | Lambda
viewport: width=device-width, initial-scale=1

[canonical-links]
https://lambda.ai/research

[document-links]
/
1-Click Clusters: /1-click-clusters
1-Click Clusters™: /1-click-clusters
AI factories: /ai-infrastructure
AI infrastructure: /ai-infrastructure
About: /about
Apply for a grant: https://form.typeform.com/to/Wn8YWMIj
Apply now: https://form.typeform.com/to/Wn8YWMIj
Blog: /blog
Brand guidelines: /brand-guidelines
CVPR 2026 Fractured Object Recovery Reassembles what's left. Generates what isn't.: https://cvpr.thecvf.com/virtual/2026/poster/40153
CVPR 2026 PixARMesh Mesh-native autoregression for whole scenes from a single view, token-by-token.: https://mlpc-ucsd.github.io/PixARMesh/
Careers: /careers
Compare: /cloud
Create account: /sign-up
Customer stories: /customer-stories
Documentation: https://docs.lambda.ai
Enterprise: /enterprise
GPU benchmarks: /gpu-benchmarks
Get started: https://news.lambda.ai/news/today
Government: /government
ICLR 2026 AgentFlow Train the agent inside its own loop. 7B beats GPT-4o on search, math, and science.: https://agentflow.stanford.edu/
ICLR 2026 Align Your Structures for Molecular Dynamics Pretrains on static molecular structures, stitches them into dynamics trajectories, bypassing simulation data scarcity.: https://arxiv.org/abs/2604.03911
ICLR 2026 EdiVal-Agent Object-level grading of multi-turn image editing, exposing where today's best editors silently break.: https://arxiv.org/abs/2509.13399
ICLR 2026 Exponent-Concentrated FP8 Model weight exponents cluster into 2–3 bits of entropy. Lossless FP8 compression, up to 177% faster inference.: https://arxiv.org/abs/2510.02676
ICLR 2026 LLM Unlearning Reframed as Retrieval Smarter data selection pushes the forget-vs-retain frontier past oracle sampling.: https://openreview.net/forum?id=Xn6EnJZghu
ICLR 2026 Latent Particle World Models Discovers objects and masks from raw video, predicts what happens next, no supervision.: https://sites.google.com/view/lpwm/home
ICLR 2026 Multi-Agent Social Interactions LLMs fold under peer pressure. Our benchmark and RL recipe help smaller models hold their ground.: https://openreview.net/forum?id=gF31wuYdk7
ICLR 2026 OffTopicEval Give an LLM a job and clear boundaries. It still answers off-topic questions, almost every time.: https://arxiv.org/abs/2509.26495
ICLR 2026 Principled RL for Diffusion LLMs Token-level RL doesn't fit diffusion LLMs. Treat the whole sequence as one action, 20–40 point gains.: https://arxiv.org/abs/2512.03759
ICLR 2026 TangoFlux 515M params, 30s of studio audio in under 4 seconds. Aligned by ranking its own outputs.: https://tangoflux.github.io/
ICLR 2026 Video Native Sparse Attention Learnable sparse attention for video. 3.6% attention budget at 128K tokens, accuracy still improves.: https://arxiv.org/abs/2510.02295
ICLR 2026 Zeroth-Order Federated LLM Fine-Tuning Sparse updates make syncing cheap enough to go frequent, neutralizing non-IID drift.: https://arxiv.org/abs/2506.03337
Instances: /instances
Instances: /sign-up
Investors: /investors
LLM index: https://lambda.ai/inference-models
Lambda Stack: /lambda-stack-deep-learning-software
Leadership: /leadership
Learn more: /gpu-benchmarks
Learn more: https://github.com/LambdaLabsML/diffusion-from-scratch
Learn more: https://lambdalabsml.github.io/Open-Sora/introduction/
Learn more: https://mlcommons.org/benchmarks/training/
Learn more: https://www.linkedin.com/pulse/architects-secure-runner-up-arc-prize-2025-lambda-cloud-rpakc/
Log in: /login
NeurIPS 2025 BLEUBERI Using simple BLEU scores as feedback on hard instructions can train instruction-following models that rival those tuned with expensive learned rewards.: https://github.com/lilakk/BLEUBERI
NeurIPS 2025 Bifrost-1 Aligns VLMs with diffusion models through shared CLIP patch embeddings, enabling controllable high-quality generation while preserving reasoning.: https://bifrost-1.github.io/
NeurIPS 2025 OverLayBench A dataset that stress-tests layout-to-image models on heavily overlapping scenes, exposing current failures and offering an improved baseline.: https://neurips.cc/virtual/2025/poster/121763
NeurIPS 2025 Scaling laws for diffusion models Diffusion beats autoregressive in data-constrained settings.: https://blog.ml.cmu.edu/2025/09/22/diffusion-beats-autoregressive-in-data-constrained-settings/
NeurIPS 2025 SpatialReasoner Builds an explicit 3D scene and reasons over it step by step, boosting accuracy and generalization on 3D spatial QA benchmarks.: https://spatial-reasoner.github.io/
NeurIPS 2025 Tensor decomposition for force-field prediction Replaces heavy tensor operations in molecular force-field models with low-rank approximation, reducing compute while keeping accuracy.: https://arxiv.org/pdf/2507.01131
Orchestration: /orchestration
Partners: /partners
Pricing: /pricing
Privacy Policy: /legal/privacy-policy
Read more: https://arxiv.org/abs/2504.18116
Read more: https://arxiv.org/abs/2505.01481
Read more: https://arxiv.org/abs/2507.04590
Read more: https://arxiv.org/pdf/2502.19587
Read more: https://arxiv.org/pdf/2503.09532
Research: /research
See benchmarks: /llm-benchmarks-leaderboard
Startups and Researchers: /instances
Startups and researchers: /instances
Superclusters: /superclusters
Superintelligence: /superintelligence
Support: /support
Talk to our team: /talk-to-our-team
Terms of Service: /legal/terms-of-service
Trust and security: /trust
Vera Rubin: https://lambda.ai/vr-200
https://www.linkedin.com/company/lambda-cloud/
https://www.youtube.com/@lambda-ai/
https://x.com/LambdaAPI

[content]
AI Research & Publications on AI Computing Platform | Lambda
Skip to main content
AI factories
Products
Supercomputers
Superclusters
1-Click Clusters™
Instances
Compare
For every mission
Superintelligence
Enterprise
Government
Startups and Researchers
Foundations
AI factories
Vera Rubin
Orchestration
Lambda Stack
Trust and security
Docs
Customer stories
Documentation
Blog
Research
Pricing
Company
Inside Lambda
About
Careers
Leadership
Investors
Resources
Research
LLM index
GPU benchmarks
Customer stories
Support
Blog
Partners
Log in
Get started
Create account
Talk to our team
AI factories
Products
Supercomputers
Superclusters
1-Click Clusters™
Instances
Compare
For every mission
Superintelligence
Enterprise
Government
Startups and Researchers
Foundations
AI factories
Vera Rubin
Orchestration
Lambda Stack
Trust and security
Docs
Customer stories
Documentation
Blog
Research
Pricing
Company
Inside Lambda
About
Careers
Leadership
Investors
Resources
Research
LLM index
GPU benchmarks
Customer stories
Support
Blog
Partners
Log in
Get started
Create account
Talk to our team
Shaping the future of AI development
Compute, community, and cutting-edge research for AI developers defining what's next
Apply for a grant
Building the future of AI through research and mentorship
Publications in 2026
CVPR 2026
PixARMesh
Mesh-native autoregression for whole scenes from a single view, token-by-token.
CVPR 2026
Fractured Object Recovery
Reassembles what's left. Generates what isn't.
ICLR 2026
AgentFlow
Train the agent inside its own loop. 7B beats GPT-4o on search, math, and science.
ICLR 2026
Latent Particle World Models
Discovers objects and masks from raw video, predicts what happens next, no supervision.
ICLR 2026
Multi-Agent Social Interactions
LLMs fold under peer pressure. Our benchmark and RL recipe help smaller models hold their ground.
ICLR 2026
EdiVal-Agent
Object-level grading of multi-turn image editing, exposing where today's best editors silently break.
ICLR 2026
OffTopicEval
Give an LLM a job and clear boundaries. It still answers off-topic questions, almost every time.
ICLR 2026
Zeroth-Order Federated LLM Fine-Tuning
Sparse updates make syncing cheap enough to go frequent, neutralizing non-IID drift.
ICLR 2026
LLM Unlearning Reframed as Retrieval
Smarter data selection pushes the forget-vs-retain frontier past oracle sampling.
ICLR 2026
Principled RL for Diffusion LLMs
Token-level RL doesn't fit diffusion LLMs. Treat the whole sequence as one action, 20–40 point gains.
ICLR 2026
Align Your Structures for Molecular Dynamics
Pretrains on static molecular structures, stitches them into dynamics trajectories, bypassing simulation data scarcity.
ICLR 2026
TangoFlux
515M params, 30s of studio audio in under 4 seconds. Aligned by ranking its own outputs.
ICLR 2026
Video Native Sparse Attention
Learnable sparse attention for video. 3.6% attention budget at 128K tokens, accuracy still improves.
ICLR 2026
Exponent-Concentrated FP8
Model weight exponents cluster into 2–3 bits of entropy. Lossless FP8 compression, up to 177% faster inference.
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%.
Learn more
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.
See benchmarks
ML Times
Your go-to source for the latest in the field, curated by AI. Sift through the excess. Make every word count.
Get started
Best practices and system insights
01
Diffusion from scratch
A guide for diffusion models implemented in a single PyTorch script.
Learn more
02
Text2Video pretraining
Lessons learned from training a text-to-video model with hundreds of GPUs.
Learn more
03
GPU benchmarks
Throughput GPU benchmarks for training deep neural networks.
Learn more
04
MLCommon benchmark
Time-to-solution benchmark for training foundation models on clusters.
Learn more
Recognized by scholars and industry peers
NeurIPS 2025
Scaling laws for diffusion models
Diffusion beats autoregressive in data-constrained settings.
NeurIPS 2025
SpatialReasoner
Builds an explicit 3D scene and reasons over it step by step, boosting accuracy and generalization on 3D spatial QA benchmarks.
NeurIPS 2025
Tensor decomposition for force-field prediction
Replaces heavy tensor operations in molecular force-field models with low-rank approximation, reducing compute while keeping accuracy.
NeurIPS 2025
Bifrost-1
Aligns VLMs with diffusion models through shared CLIP patch embeddings, enabling controllable high-quality generation while preserving reasoning.
NeurIPS 2025
BLEUBERI
Using simple BLEU scores as feedback on hard instructions can train instruction-following models that rival those tuned with expensive learned rewards.
NeurIPS 2025
OverLayBench
A dataset that stress-tests layout-to-image models on heavily overlapping scenes, exposing current failures and offering an improved baseline.
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.
01
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
Read more
02
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
Read more
03
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
Read more
04
Think, prune, train, improve
Scaling reasoning without scaling models
Caia Costello, Simon Guo, Anna Goldie, and Azalia Mirhoseini — ICLR 2025 workshop
Read more
05
NeoBERT
A next-generation BERT
Lola Le Breton, Quentin Fournier, Mariam El Mezouar, and Sarath Chandar — TMLR 2025
Read more
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.
Apply now
FOOTER
AI FACTORIES
For every mission
Superintelligence
Enterprise
Government
Startups and researchers
Foundations
AI infrastructure
Trust and security
Customer stories
Products
Products
Superclusters
1-Click Clusters
Instances
Features
AI infrastructure
Orchestration
Lambda Stack
Trust and security
Docs
Documentation
Blog
Research
Company
Inside Lambda
About
Careers
Leadership
Investors
Resources
Research
Customer stories
Blog
Partners
Brand guidelines
Privacy Policy
Terms of Service
Cookie preferences
Cookie management is unavailable because your browser is blocking third-party scripts.
©
2026
Lambda. All rights reserved.
