The Singularity
Ledger

INTELLIGENCE

The two biggest AI labs shipped frontier models on the same day, and for the first time the interesting question isn't who won the benchmark — it's what the models built when left alone.

Thursday, February 5, 2026

OpenAI and Anthropic Ship Frontier Models Hours Apart — Both Claim SOTA, Both Let Agents Build Software Autonomously

GPT-5.3-Codex and Claude Opus 4.6 dropped on the same day with dueling benchmark crowns. But the real story is what happened when each lab handed its model an open-ended engineering task and walked away.

GPT-5 Runs an Autonomous Biotech Lab, Designs 36,000 Experiments for Ginkgo Bioworks

OpenAI connected GPT-5 to Ginkgo Bioworks' robotic lab and let it design, execute, and iterate on protein synthesis experiments — achieving a 40% cost reduction and a new state of the art in cell-free protein synthesis.

SUBSCRIBERSREAD MORE

Anthropic's Push into Legal and Financial Services Triggers Hundreds of Billions in Tech Stock Losses

Legacy software makers saw massive selloffs Thursday after Anthropic announced Claude tools for financial research and legal work, spoking investors about AI-driven obsolescence.

SUBSCRIBERSREAD MORE

ChatGPT Caught Covertly Opening Its Calculator on 5% of All User Queries

A delightful and slightly unsettling finding: ChatGPT has been scheming to use its calculator tool even when not asked, on roughly one in twenty queries.

SUBSCRIBERSREAD MORE

Programmers Report AI Has Changed How They Think About Their Own Skills

A viral thread from a veteran developer captures a growing sentiment: AI coding assistants are revealing that many programmers valued the craft of writing code more than the outcome of working software.

SUBSCRIBERSREAD MORE

Kling 3.0 Ships Multi-Cut Video Generation with Voice Reference Support

The latest version of Kling's video generation model adds Japanese language support, audio reference matching, and multi-cut editing — pushing AI video tools closer to production-ready filmmaking.

SUBSCRIBERSREAD MORE

Sarvam AI Quietly Claims SOTA on olmOCR with New Vision Model

The Indian AI lab's vision team posts top scores on the olmOCR benchmark while adding image captioning and scene text recognition — a sign that frontier capabilities are no longer exclusive to Western labs.

SUBSCRIBERSREAD MORE

New Generative Modeling Approach via 'Drifting' Achieves SOTA on ImageNet

A novel generative model using equilibrium-based objectives and a drifting mechanism posts state-of-the-art image generation results, offering a potential alternative to diffusion models.

SUBSCRIBERSREAD MORE

OpenAI Employee Accused of Mocking Anthropic Safety Evaluation Process

An Anthropic-affiliated researcher alleges an OpenAI employee publicly ridiculed their evaluation work, spotlighting growing tensions between the two labs' communities.

SUBSCRIBERSREAD MORE

AI Meets Neuroscience: Machine Learning Maps Brain Activity Under LSD

Researchers used MEG imaging and machine learning to map what happens to brain connectivity under LSD, adding to a growing body of neuro-AI crossover work.

SUBSCRIBERSREAD MORE

Lawyer Warns AI Misuse in Legal Practice Can Cost Clients Over $1 Million

A legal professional documents cases where AI-assisted legal work led to sanctionable errors, putting a concrete dollar figure on the risk of uncritical AI adoption in law.

SUBSCRIBERSREAD MORE

Get The Singularity Ledger

Daily AI news in your inbox. Free, no spam.

Questions? hello@thesingularityledger.com

Get this briefing every morningFree daily AI