On this day: May 7

8 records from this date, 2023 to 2026, gathered year by year — the most recent year first.

2026

This year · 5 records
Stop Comparing LLM Agents Without Disclosing the HarnessZhang et al. (arXiv:2605.23950)

Harness-induced variance was 7.80× model-induced variance, and rankings flipped in 6 of 9 model pairs

GPT-Realtime-2 / -Translate / -WhisperOpenAI

Brought reasoning effort and parallel tool calls to voice models and raised context from 32K to 128K. Voice became an environment for running agents

MultimodalOfficial
Moonshot AI raises $2 billion at a $20 billion valuationMoonshot AI

Raised $3.9 billion over 6 months. A case showing that open-weight distribution works commercially

Google DeepMind reports AlphaEvolve's first-year resultsGoogle DeepMind

Real progress in domains that can be checked automatically (Google's own measurement)

EU Council and Parliament provisionally agree on the Digital Omnibus (AI)Council of the EU, European Parliament

The moment the effective postponement of the AI Act's high-risk obligations was decided

2025

1 year ago · 1 record
PyTorch Foundation takes in vLLM as a hosted projectPyTorch Foundation

The serving engine went from research project to infrastructure public good

2024

2 years ago · 1 record
DeepSeek-V2DeepSeek-AI

A third path, distinct from MQA and GQA, that compresses KV into a low-dimensional latent

2023

3 years ago · 1 record
Language Models Don't Always Say What They Think publishedMiles Turpin et al. (NYU, Cohere, Anthropic)

The starting point of the line of work challenging CoT faithfulness