Your briefing

Models

OpenRouter introduces side-by-side benchmarks for model routers

The new Router Index compares routers across six suites using quality, speed and cost to help builders decide whether to use one and which to choose.

Models

Study finds episodic and hybrid context strategies lead on long-distance clinical reasoning

An evaluation across four open-weight models found that episodic and hybrid approaches performed best overall and when supporting evidence was far from a question.

Models

Study finds post-training can weaken LLM grammar across 101 languages

An evaluation of six model families finds uneven effects from post-training, with low-resource languages facing the largest costs.

Computational tools for society’s most complex challenges

Agents

Okta outlines user identity handoff for Amazon Bedrock AgentCore tool calls

Okta says a token exchange using Cross-App Access and the Identity Assertion Authorization Grant can preserve user attribution and agent identity across tool calls.

The Agent Said It Was Done. The Database Disagreed.

Agents

Cloudflare releases Clef decision models on Workers AI

Clef and Clef-flash return structured probabilities for agent decisions, and their weights are available under Apache 2.0.

Agents

Decagon joins OpenAI’s B2B marketplace as a launch partner

The partnership gives eligible OpenAI enterprise customers a way to apply part of an existing commitment toward Decagon’s customer-facing AI agents.

All the AI agents that can live in your text messages

Apple says it’s tightening macOS ‘Full Disk Access’ controls due to new risks from AI agents

Trump unveils his new Super Intelligence Force

Tools

Browserbase Functions add runtime access to managed Secrets

Developers can store credentials in Browserbase and retrieve them from Functions at runtime for authenticated workflows.

Tools

Haystack 3.3.0 speeds up SentenceWindowRetriever and updates BM25 behavior

The release cuts Document Store lookups for SentenceWindowRetriever and changes which documents in-memory BM25 retrieval returns.

Tools

Unsloth adds desktop command palette and lower-memory LoRA training

The October 1 release also brings faster Laya decisions, shareable GGUF settings and additional fine-tuning support.

Meta wants your next gadget to be Muse-infused

Tools

Mux launches dashboard playground for testing video workflows

The Mux Robots playground lets builders run workflows on their video, adjust settings and compare results before integrating them into an application.

Tools

GitHub opens Copilot code review requests through APIs

Copilot users on eligible plans can request reviews through REST and GraphQL APIs and set an effort level for each request.

Sean Parker is rebuilding Stability AI around music

Tools

Kiro CLI 2.27.0 adds workflow controls, live Steering context and prompt commands

The update lets builders configure how V3 delegates with Workflows and adds live context references and saved-prompt slash commands.

OpenAI safety employee resigns, claiming the company’s ‘culture is broken’

It’s not AI anymore, it’s ‘super intelligence’ (according to the White House)

Infrastructure

Neon raises free-plan database storage to 1 GB per project

The higher limit applies automatically to existing projects and sits alongside a free allowance of up to 100 projects.

Amazon responds to data center backlash, says it no longer uses NDAs

Business

Fireworks raises rate limits for Small serverless models

The expanded Small tier now covers models below 600B parameters, with higher ceilings for prompt and generated tokens.

Affected by layoffs? Don’t miss this $75 deal for your TechCrunch Disrupt 2026 Expo+ Pass

Call it AI, call it Super Intelligence, only 2% of consumers are buying it