The Token Reporter

A day in AI

Tuesday, 20 October 2026

40 stories · 8 from multiple sources

▲ 2 signals · 2 sources

How Claude designed protein binders against scorpion neurotoxins

Science

anthropic.com · 26 days ago

▲ 3 signals · 3 sources

People are using AI to make their smart TVs dumb again

techcrunch.com · 27 days ago

▲ 3 signals · 3 sources

A hacker broke into John Deere and SharkNinja with an army of AI agents

theverge.com · 27 days ago

More stories

▲ 2 signals · 2 sources

DeepSeek just open-sourced its inference engine

github.com · 26 days ago

▲ 3 signals · 2 sources

Latest open artifacts (#25): DeepSeek V4.1 Flash 1017, American-made models, and new regulatory controls in the US & China

interconnects.ai · 27 days ago

▲ 2 signals · 2 sources

Mercury 3: Diffusion at scale

inceptionlabs.ai · 27 days ago

▲ 2 signals · 2 sources

Cutting our inference bill by 68% by being obsessive about optimization

blog.cloudflare.com · 27 days ago

Build a semantic layer for agentic AI on AWS with Denodo and Amazon Bedrock AgentCore

aws.amazon.com · 26 days ago

Strengthening our defenses against distillation attacks

anthropic.com · 26 days ago

Also today

OpenAI acquires Orchestra Bio

incoai/Qwen3.8-Flash-Next-DFlash2

Battle Drones - a vibe coded RTS with audio-based multiplayer

simonwillison.net

Franchisee reconciles cash in 94% less time with ChatGPT Work

Anthropic logged net loss of over $80B in first half of 2026

On AI writing

Microsoft says AI agents are like insider threats as ‘frontier models can’t be trusted’

NVIDIA Cosmos3-H-Dreams: Bringing Clinical Surgery to Life Through Generative Simulation

Google open-sources TorchTPU, a native PyTorch back end for TPUs

FairStress: A deeper evaluation of fairness in LLM decision-making

Steam Users Beg for "No AI" Filter as Vibecoders Flood Next Fest

The Programmable Data Center — Evan Conrad, SF Compute

Beyond the KV cache: Training and serving i14’s Laplace 3B linear attention model on AWS Trainium2 with native PyTorch

XHToken/Spark-X2.6-4B · Hugging Face

qev-0.8b · New SOTA tiny local decision model

Work management context for AI agents with Smartsheet MCP on AWS

From data to decisions: Conversational supply chain intelligence with Amazon Quick and Amazon Bedrock AgentCore

Fine-tuning GPT-OSS-20B for multilingual reasoning on Amazon SageMaker HyperPod

What’s the ROI of software engineers’ spending on AI?

newsletter.pragmaticengineer.com

How Octave built a secure multi-tenant AI foundation on Amazon Bedrock

Rethinking confidence in reasoning models

Agnes-AI/Agnes-3.0-Qwen (QFN finetune)

Short report from Lithuanian conference on AI Technologies in Medicine

Access OpenAI GPT models on Amazon Bedrock from Canada with cross-Region inference

Discover tools at runtime with AWS Agent Registry

MiMo 2.6 is reasoning is like the crash test dummies

Phonon is now open-source, 31M and 40M TTS models designed to run on the edge

30 Months on the Exponential (in Charts)

oneusefulthing.org

Hardly any benchmarks for Intel B65, so I got 4x and tested them. Long Context Sweeps for Flash-Next and other models

Pocket TTS which supports voice cloning is now just 1.9 million parameters

DeepSeek Dials Up the Pressure on OpenAI and Anthropic With Open-Source Inference Tool

← All editions