A day in AI

Tuesday, 6 October 2026

40 stories · 4 from multiple sources

▲ 9 signals · 4 sources

Mistral Large 4

My comment on Mistral Large 4 — Hacker News. wren6991 : The benchmark is saturated. Frontier models are tested with an armadillo in fishnet tights jaywalking on Mars. OK well I couldn't resist this one: llm -m claude-opus-5.5 'Generate an SVG of an armadillo in fishnet tights jaywalking on Mars' llm -m gpt-6.1-sol 'Generate an SVG of an armadillo in fishnet tights jaywalking on Mars' llm -m…

simonwillison.net ·

Also leading

Introducing GLM 5.3 on Amazon Bedrock▲ 3 signals · 2 sourcesaws.amazon.com·llm-mistral 0.16simonwillison.net·Advancing computer use with Ironcladopenai.com·Expanding the Cyber Verification Programanthropic.com·
▲ 5 signals · 3 sources

EmbeddingGemma 2: an open, lightweight multimodal embedding model

deepmind.google ·
▲ 3 signals · 2 sources

Qwen3.8 27B addition in words

simonwillison.net ·

More stories

datasette-atom 0.11a0

simonwillison.net ·

Qwen3.8-Flash-Next (125B) on a single Strix Halo mini PC: 44-59 tok/s with speculative decoding, ~1,400 tok/s prefill, engine is open

reddit.com ·

Atlassian and OpenAI expand partnership to turn enterprise knowledge into action

openai.com ·

Using Parseable with Datasette for OpenTelemetry traces

simonwillison.net ·

OpenAI Decisions API is in public beta

developers.openai.com ·

Claude Code’s suggested message feature: I think the real customer is the model

zohaib.cc ·

Also today

Utah to let AI examine patients and prescribe medication without human oversight

techspot.com

AI is now capable of developing its own inference hardware

github.com

Show HN: OpenChart – OSS TradingView alternative with your own AI agent

github.com

The keys to the Internet change on October 11. Are you ready?

blog.cloudflare.com

Erdosproblems.com Succumbs to the AI Onslaught

erdosproblems.com

What AI gets wrong and what failure teaches us

microsoft.com

GPT-6.1 Sol looped "leak" hints at nested models serving architecture

reddit.com

Building a context-aware AI assistant on AgentCore and OpenClaw

aws.amazon.com

Local AI ecosystem overview

reddit.com

I trained a model to be wrong 98% of the time and 96% sure about it. It took three tries.

reddit.com

Beyond synthetic testing: Capturing and replaying real database workloads at Airbnb

medium.com

I gave a 21M model a 6.4B-parameter lookup table. It matches a 114M dense model and runs with the table on an SSD (RX 9070)

reddit.com

EmbeddingGemma 2 running locally in-browser on WebGPU

reddit.com

European AI flag bearer Mistral's new open weights model is 'Le Chonk'

theregister.com

How AI decision models could change content moderation

techcrunch.com

AI computing startup Lambda to raise $4B ahead of planned IPO

techcrunch.com

The next hurdle for AI agents: getting websites to let them in

techcrunch.com

Hark releases an AI personal assistant with a focus on privacy

techcrunch.com

Mirror Particle is building a ‘world model’ of human behavior

techcrunch.com

We can’t just change the definition of ‘recording’

theverge.com

Anthropic is giving startups a free year of Claude Team and $1,000 in credits

techcrunch.com

Scrimshaw Jukebox

simonwillison.net

Unlocking Earth AI’s planetary geospatial foundation models for global public health

research.google

Show HN: Jotbus – a shared encrypted scratchpad for coding agents

jotbus.com

Oracle Triggered the Implosion of the AI Bubble

medium.com

NTS: Authenticated Time at Meta

engineering.fb.com

The Cyber Risk Discourse is Broken

interconnects.ai

← All editions