AI21 Labs

AI21 Labs launches Teel, a Python runtime for faster agent workflows
AI21 Labs has introduced Teel, an open source Python runtime designed to overlap independent, IO-bound calls in ordinary sequential Python code. The company says Teel…

AI21 says post-training pushed Nemotron 3.5 Super past frontier quality
AI21 says it has post-trained NVIDIA Nemotron 3.5 Super VL to outperform Claude Opus 5.5 on Deel’s global payroll benchmark, while costing 50x less per 1M tokens. The…

AI21 and Alice post-train Nemotron 3.5 Super VL to beat GPT-6.1 Sol on attack resilience
AI21 Labs says its Forward-Deployed Researchers, working with AI safety company Alice, post-trained NVIDIA's Nemotron 3.5 Super VL in three weeks to resist attacks such…

AI21 Labs: Standard Instruction-Following RL Taught a Model to Answer 'China'
AI21 Labs says a model trained with reinforcement learning on the open IFTrain dataset learned to satisfy constraints while skipping the task, answering a request for an…

AI21 Cuts GitLab Duo Code Review Cost by 98% by Tuning Its Harness
AI21 Labs says it cut the cost of GitLab Duo Code Review by 98% on Martian's Code Review Bench without losing review quality. According to the company, all of the saving…

AI21 Post-Trains Nvidia Nemotron 3.5 Super VL for Gong, Deel and Alice
AI21 Labs says its Forward-Deployed Researchers post-trained NVIDIA Nemotron 3.5 Super VL with Gong, Deel and Alice, producing specialized models that match or surpass…

AI21 Labs: Public coding benchmarks miss how developers really use agents
AI21 Labs says public coding benchmarks such as SWE-bench, HumanEval and Spider give a skewed picture of how developers use AI agents. After classifying seven weeks of…

AI21 Labs: Check Whether an Agent Can Verify Its Answer Before Scaling
AI21 Labs argues that engineers should ask one question before adding compute to an AI agent: can the agent recognize when it has the right answer? According to the…

AI21 Labs Replaces Slack-Style GPU Haggling With Kueue on a 10,000-GPU GKE Cluster
AI21 Labs has described how it moved a shared cluster of around 10,000 GPUs from manual coordination in a messaging channel to Kueue, an open-source Kubernetes-native…