AI News

Reality check

Jev Reality Check: TypeSafe's Model Stays Closed, but Open Jev-Class Models Now Run Locally

Jev itself is still closed, but open Jev-class models now run on your own hardware. What works in Ollama, and what fits your GPU.

Sep 25, 202610 min read

Claude

Is Claude Opus 5.5 Cheaper? What the Price Cut Changes for API Users, Subscribers, and Local AI

Claude Opus 5.5 is 20% cheaper per token, about 40% per workload. Subscribers get bigger limits, not a lower bill. And the catches nobody mentions.

Sep 22, 202610 min read

Fable

Claude Fable 5.1 Ships With a Text Watermark: What It Proves, Whether You Can Remove It, and the Self-Hosted Way Around It

Fable 5.1 is the first Claude model that watermarks its text. What the mark proves, whether it can be removed, and why self-hosting is the clean answer.

Sep 1, 202610 min read

GLM

GLM-5.3: Weights in Two Weeks, and What Z.ai's Bug Ledger Means for Your Network

GLM-5.3 is live; open weights land in about two weeks. What Z.ai's 2,436-finding bug ledger means for the software running your network.

Aug 14, 20269 min read

DeepSeek

DeepSeek V4: Price, Specs, and Status After V4-Pro-0813 (October 2026)

V4-Pro left preview on August 13, API prices went up, and the Flash API now serves V4.1-Flash-1017. The dated ledger, verified October 18.

Aug 4, 202610 min read

Qwen

Qwen3.8's 2.4T Open-Weights Promise: Now With a Date, Still Without a License

Alibaba teased a 2.4T open-weight Qwen3.8 three days after Kimi K3. Now there's a date and pricing. The updated ledger of verified vs claimed.

Aug 3, 202610 min read

DeepSeek

DeepSeek V4-Flash 0731: The Upgrade Everyone Can Use and Nobody Can Download

DeepSeek upgraded the V4-Flash API to build 0731 with big agent gains. The open weights are still the April preview. Here is the honest split.

Jul 31, 20269 min read

Kimi

Can You Run Kimi K3.1 Locally? Same 2.8T K3 Shape, Weights Due Nov 4

Kimi K3.1 Pro is live in the API; weights land by Nov 4. Same 2.8T/104B shape as K3, so the home-hardware answer has not moved. What to run instead.

Jul 30, 202611 min read

Claude

Are Your Shared Claude Chats Public? How to Check and Revoke Them

Shared Claude chats and artifacts appeared in Google results. What actually happened, why it keeps happening, and how to audit and revoke your shares.

Jul 27, 202610 min read

Claude

Claude Opus 5 Ignoring Your Instructions? Your Old Prompts Are Probably Why

Verbose output, scope creep, early stopping, token burn. Anthropic documented all four the day Opus 5 shipped, along with the prompts to delete.

Jul 24, 202610 min read

Fable

Claude Opus 5 or Fable 5: Which Model to Use, and What Each One Will Refuse

Anthropic's cheaper model is the less restricted one on cybersecurity and biology. A plain decision guide to Opus 5, Fable 5, Sonnet 5, and Kimi K3.

Jul 24, 202610 min read

Security

The Hugging Face Breach Had a Twist: The AI Guardrails Blocked the Defenders

OpenAI's escaped eval models breached Hugging Face. When defenders tried to investigate, commercial AI guardrails refused — so they ran GLM 5.2 locally.

Jul 21, 202610 min read

Gemini

Can You Run Gemini 3.6 Flash Locally? No — Here's What to Run Instead

Gemini 3.6 Flash launched July 21 with no local option. The honest answer, why Google's efficiency pitch favors local AI, and the Gemma 4 path instead.

Jul 21, 20269 min read

Kimi

Kimi K3 vs Claude Fable 5 vs GPT-5.6 Sol: What the Benchmarks Actually Show

Moonshot's Kimi K3 beat Claude Fable 5 and GPT-5.6 Sol on one major leaderboard and lost on others. A plain-language look at what the numbers really say.

Jul 17, 20265 min read

How Long Would It Take to Download Kimi K3 on Your Internet Plan?

Moonshot AI's Kimi K3 is the largest open AI model ever released, and downloading it is a real test of your internet plan. We break down download times by speed tier and explain why a single download can exceed the Cox and...

Jul 17, 20265 min read

Kimi

Is Kimi K3 Free? What You Actually Pay With

Everyone says try Kimi K3 free. We read the privacy policy: training on your content, no documented opt-out, and a storage jurisdiction it never names.

Jul 16, 202610 min read

Fable

Is Fable 5 Nerfed? What the Benchmark Crash Actually Measures

One benchmark says the returned Fable 5 lost 60 points on debugging. The honest read: the gate got tighter, not the brain, and nobody can verify either.

Jul 2, 202610 min read

Fable

Claude Fable 5 Is Back After US Lifts Export Controls

The US government lifted export controls on Anthropic's Claude Fable 5 and Mythos 5 on June 30, 2026. Fable 5 returns to users worldwide on July 1. Here is a clear, non-technical rundown of what happened, why the models were pulled, and...

Jul 1, 20264 min read

Claude

Claude Sonnet 5: Price, Benchmarks, and What It Doesn't Replace

Claude Sonnet 5 launched today at a fraction of Opus 4.8's price. Here's what's new, what it costs, and why it's the real ceiling for now.

Jun 30, 20268 min read

Policy

It Was Never About Dario: The Deep History of Government AI Control

Pulling Claude offline and gating GPT-5.6 was not a surprise. The 70-year pattern behind AI's permission layer, and the local hedge that beats it.

Jun 27, 202610 min read

Policy

When the Government Decides Who Gets AI: The GPT-5.6 Signal

For the first time, Washington is gating who can run a frontier AI model. What the GPT-5.6 and Fable 5 moves mean, and the hedge that still works.

Jun 26, 202610 min read