not anapp
Latest newsOpenAIAnthropicGoogleGrokMetaAI VideoOpen SourcePricing

The latest in AI

News + context
Table: GPT-6.1 Terra passed all 186 checks at about 0.93 cents per job on medium effort and 0.51 cents on low; GPT-6.1 Sol 1.25, GPT-6 Luna 0.12 and Claude Haiku 5.5 0.19 cents
The big picture

GPT-6.1 Terra tested: every check passed, for half a cent a job on low effort

We gave OpenAI's new GPT-6.1 Terra our secret coding test. It passed every check, about half a cent a job on low effort, but Claude Haiku 5.5 and GPT-6 Luna did it for less.

Marvin Smit· 6 min read
Pricing

ChatGPT Plus vs Pro: is Pro worth it? We ran 54 coding jobs to find out

ChatGPT Pro now costs $100, $200 or $500. We ran 54 coding jobs in Codex on a Pro 200 plan: all passed, and the weekly counter moved from 7% to 8%. Here is what fits.

Marvin Smit· 8 min read
Open Source

PHP2Shell: AI helped find two critical PHP flaws. What to do now

Two critical PHP flaws, PHP2Shell, let attackers run code on web servers without logging in. Wiz found them with help from AI. Who is at risk and how to fix it.

Marvin Smit· 8 min read

More to follow

Across the AI landscape
Open Source

MiniMax M3.1 Flash tested: cheap per token, but not per coding job

Marvin Smit· 9 min read
Open Source

LM Studio vs Ollama: we ran a local AI model on a normal laptop

Marvin Smit· 11 min read
Open Source

PPT Master tested: free AI PowerPoint slides from Claude Code

Marvin Smit· 9 min read
M
Meta

Muse Spark Open tested: Meta's open model, its licence and Muse Wonder

Marvin Smit· 9 min read
G
Grok

Grok Bot vs Claude Cowork: we gave both a family's inbox

Marvin Smit· 10 min read
G
Grok

The X Subscription: Grok, Grok Bot and Cursor in one plan

Marvin Smit· 7 min read