Search papers, labs, and topics across Lattice.
4
3
10
6
Synthesized software environments unlock a 50% boost in CUA performance on unseen benchmarks, proving scalable supervision is key.
The DeepSeek-V4.1-Flash model, a multimodal Mixture-of-Experts model with 552B backbone parameters and support for contexts of up to one million tokens, is introduced, substantially improving cost efficiency for agentic workloads and pushing the limits of KV cache compression.
ARENA uncovers vulnerabilities in large audio-language models that traditional text-based red-teaming methods miss, achieving near-perfect safety metrics across multiple systems.
ABSeeker's innovative credit assignment method allows it to achieve performance levels comparable to much larger models, redefining expectations for long-horizon search agents.
