OpenAI AI News, Models and Product Updates
Track latest Openai AI news, launches, research, and ecosystem moves.
Openai news, model releases, product launches, research updates, and major announcements in one place.
Latest Articles
- Separating Common and Unique Directions for Data-Free Model Merging — CUSD, a data-free pre-merging step that splits task vectors into shared and task-specific directions, improved 42 of 44 model-operator combinations and gained up to 22.5% over standard arithmetic model merging.
- TAPES: Task-Aligned Joint Predictive Pretraining for Transaction Event Sequences — TAPES picks pretraining targets for transaction-event sequences by their measured downstream value, then pretrains jointly with a shared discrete latent, beating baselines on a real transaction-risk task and public finance benchmarks.
- Measuring Model-Specific Alignment Drift in LLM-Based Opinion Dynamics — LLM agents in opinion-dynamics simulations lose assigned stances through interaction in model-specific ways, with gpt-4.1 keeping clusters apart while gpt-5.6-sol and claude-opus-5 abandon them, per a new study.
- Failure-Driven Workflow Refinement — CE-Graph refines LLM workflows by modeling where they fail rather than optimizing a single score, editing workflow graphs to reduce the 'expected failure mass' found in a pool of counterexamples.
- OSGR: Observation-Support Graph Reasoning for Earth Observation Question Answering — OSGR, a two-stage graph reasoning framework for Earth observation question answering, lifts GPT-4o accuracy from 49.3% to 63.7% on Earth-Bench and DisasterM3-Reason while cutting execution steps.
- Large Language Models for Forecasting of Electricity Price Spikes — A preregistered-style test on Australia's electricity market found LLM-read market notices did not reliably improve short-horizon price-spike forecasts, a useful negative result for LLM feature extraction.
- Navigating Runtime Dynamism in Semantic Query Serving — SeqoFlow is a runtime for LLM-powered semantic queries that adapts query plans mid-execution and schedules concurrent queries together, cutting end-to-end latency compared with existing semantic query systems.
- GC-MoE: Overlapping Expert-Parallel Token Dispatch with FP8/FP4 Grouped GEMM via CUDA Green Contexts — GC-MoE uses CUDA Green Contexts to split GPUs into separate communication and compute partitions for MoE inference, speeding DeepSeek-V4.1-Flash decode by about 1.19x on 8 H20s and beating DeepGEMM Mega MoE.
- Spectral Quantum Memory for Implicit Neural Representations — A NeurIPS 2026 paper introduces the Spectral Quantum Memory Model, a quantum implicit neural representation where layers share a coherent memory register, improving over compact baselines on audio and image reconstruction.
- Data centers are turning to old jet engines as the wait for power grows — AI data center builders stuck in multi-year grid and turbine queues are installing rebuilt jet engines burning natural gas, showing how far power supply now lags the AI buildout.