What actually shipped in agent engineering, pulled from the labs, arXiv and Hacker News.
See who we followToyBlocks is an open simulation framework for studying robot learning through automated task generation, training demonstrations, and evaluation in a controlled tabletop world.
Ai2 and Google developed NeuralGCM+, an open-source climate model combining physics and AI for atmosphere and ocean simulation.
StateBench reveals language models struggle tracking state changes in long documents; targeted training helps recurrent and hybrid architectures.
Nextcloud selected Olmo 3 for fully open, self-hosted AI with highest ethical rating on consumer hardware.
Researchers adapted olmOCR 2 for searchable historical archives and archaeological handwriting.
Old Ideas, Novel Problems study shows LLM novelty judgments vary significantly with minor evaluation setup changes.
PLANTEUR extends crop-mapping benchmark to six tropical regions for evaluating geospatial models.
OlmoEarth v1.3 provides state-of-the-art satellite embeddings and browser-based explorer for global analysis.
FairStress benchmark shows LLMs exhibit demographic differences in how they defend decisions despite appearing fair overall.
HiRO-ACEv2 simulates extreme weather at 3 km resolution for 1/300th the cost of X-SHiELD.
MIT researcher used open-source Molmo 2 to build Molmo2Fish, a model tracking salmon in sonar footage.
Ai2 presented advances in open language models, training infrastructure, and AI for science at COLM 2026.
Byte-level language model technique published in Nature generalizes across model families.
AI2 released AstaBrief, an 8B open-weights model for generating cited scientific reports available in Fast mode or for local deployment.
Olmo-core 3 provides open training infrastructure for efficiently scaling mixture-of-experts models to trillion parameters.
Crowdsourced game testing of Olmo 3 revealed model vulnerabilities and behavior exploits.
Goodfire used Ai2's open post-training stack to trace unwanted LLM behavior to individual training examples.
BenchMIRT audits LLM benchmarks question by question to reveal measured capabilities.
Ai2 and Providence Swedish Cancer Institute expanded collaboration after AutoDiscovery validated a new immune signal in breast cancer.
Thai researchers adapted Ai2's Dolma toolkit to create Mangosteen, a 47-billion-token Thai language corpus for improved model performance.
One pattern, one tradeoff, one production failure story. A short weekly briefing for people building agentic systems.
Weekly email, one-click unsubscribe. We only use your address to send the briefing.
