Artificial Intelligence

Artificial Analysis

Independent model benchmarks & pricing

LMArena

Crowd-sourced model rankings

Epoch AI

Compute & capability trends

The Decoder

Daily AI news

Claude Code docs

Official guide to the Claude Code CLI & agents

Agent SDK docs

Build your own agents on the Claude platform

Tool use

Give Claude tools & function calling

Prompt caching

Cache context to cut cost & latency

Models overview

Opus, Sonnet & Haiku — what each is for

Pricing

Current per-model token pricing

Anthropic Engineering

First-party blog on building agents in practice

Anthropic News

Model launches & official announcements

Claude Code release notes

The running changelog for the CLI

Claude Code product

What Claude Code is and what it does

Simon Willison

Independent, hands-on notes on Claude & LLMs

Redeploying Fable 5

Fable & Mythos access

The June 12 suspension statement, since updated through the restoration.

Claude Sonnet 5

The June 30 launch announcement for the new budget model.

Fable 5 & Mythos 5 docs

The platform documentation for both Mythos-class models.

CNBC — controls lifted

June 30: the administration lifts export controls on Fable 5 and Mythos 5.

CNBC — the suspension

June 12: access disabled worldwide to comply with the directive.

The Hacker News

The restoration and the new cyber-jailbreak safeguards.

Tom’s Hardware

Fable 5 restored as the US lifts export controls.

BleepingComputer

The restoration timeline ahead of the July 1 return.

VentureBeat — the return

Where enterprises can access Fable 5 after the global restoration.

VentureBeat — Sonnet 5

Sonnet 5 launches at a steep discount to the top model.

TechCrunch

Sonnet 5 positioned as a cheaper way to run agents.

PCWorld

Subscriber backlash over the shortened, half-limit return window.

9to5Google

Fable 5 returns to Claude apps and the model picker.

9to5Mac

Fable 5 cleared to return as the restriction is lifted.

MakeUseOf

The redeployment — and its caveats.

Axios

The administration lifts the Fable 5 restrictions — the policy angle.

Washington Post

“Fable is back” — the July 1 AI & Tech Brief.

Forbes — the runway

June 29: the return telegraphed days ahead (updated after it landed).

TechBuzz

Fable 5 re-enabled on AWS Bedrock after the ban lift.

MarketScale

What the 19-day shutdown taught enterprises about AI as infrastructure.

Forbes

What happened: the June 16 explainer on the export-control order.

Vellum — benchmark breakdown

The clearest side-by-side: SWE-Bench Pro 80.3%, FrontierCode Diamond 29.3%, with the comparison field. Our July 9 numbers trace here.

Anthropic — Redeploying Fable 5

Primary source for the July return, the stricter safety classifier (>99% block of the reported bypass), and the false-positive tradeoff.

BenchLM — live leaderboard

A running, independent benchmark aggregate for Fable 5 — useful for tracking where it sits as new evals land.

CodeRabbit — model review

A code-review vendor's own eval; notable for a balanced read on where the model is strong vs merely good.

MindStudio — agentic coding, real-world

Beyond benchmarks: how Fable 5 behaves on long-running agent tasks — tool efficiency, persistence, self-correction.

Lenny’s Newsletter — right and very wrong

A product-leader review that credits the wins and is candid about the failure modes — a good skeptic's counterweight.

DataCamp — a Mythos-class model you can use

A practitioner-facing explainer of what Mythos-class means and how to actually put Fable 5 to work.

DigitalApplied — the July 12 extension

Precise timeline on the window extension and the credits-pricing switch; our lead-story dates trace here.