Aric Mitchell

I run mAIn Street newsletter (Tue-Fri) and serve as comms strategist for the @FortSmithPD. Follow me for tips, news, and insights on both AI and comms strategy.

[Aric Mitchell @AricMitchell]

@Alex_FF Nailed it!

[Aric Mitchell @AricMitchell]

@ericlu they must have fixed it because Sol Light was able to read the message i just sent it. https://t.co/SbyzzDJeMy

[Aric Mitchell @AricMitchell]

@OpenAI But where the hell is it? I don't see squat in my $200/mo account. Would be nice if you could speed up the rollout, considering you've had weeks to prepare.

[Aric Mitchell @AricMitchell]

https://t.co/CNjiyZOLE3

[Aric Mitchell @AricMitchell]

I Don’t Need AI. Neither Do You.. And That’s Exactly What You Need to… | by mAIn Street | ILLUMINATION | Jun, 2026 | Medium

[Aric Mitchell @AricMitchell]

Everything sucks nowadays.

Okay, not everything, but scroll your feed for five minutes and try to tell me I’m wrong. The writing is fine, the images are fine, the videos are fine, and somehow all of that “fine” has become the actual problem.

Here’s the trap we’ve walked into.
https://t.co/QXWgGpelyh

[Aric Mitchell @AricMitchell]

Excellent pricing-intelligence guide from @ArtificialAnlys

[Artificial Analysis @ArtificialAnlys]

Announcing Artificial Analysis Intelligence Index v4.1: a shift toward agentic workloads, featuring upgraded benchmarks and new per-task metrics

The Artificial Analysis Intelligence Index is our synthesis metric for assessing model intelligence and tracking AI progress. v4.1 marks a broader shift toward agentic workloads, with three main changes:

Updated and reweighted evaluations toward agentic tasks:
1. We upgraded three evaluations, removed one, and reweighted the Intelligence Index:
➤ Upgraded Terminal-Bench Hard to Terminal-Bench 2.1 and τ²-Bench Telecom to τ³-Bench Banking. Both move to newer, more robust task sets with harder, more realistic agentic scenarios that better separate frontier models
➤ Upgraded GDPval-AA to GDPval-AA v2. The upgrade re-baselines Elo to human performance at 1000, introduces a rotating panel of frontier-model judges, and raises the turn limit from 100 to 250 for longer-horizon agent trajectories
➤ Removed IFBench due to saturation. The benchmark no longer distinguishes frontier models sufficiently, so we have removed it from the Intelligence Index. We will continue to run it and publish results on new model releases

2. Cost per Task, Time per Task, and Tokens per Task:
Three new per-task metrics, reported for every model and based on the Intelligence Index. We take the total cost, total time, and total output tokens for a model to run the Intelligence Index and divide by the number of tasks across its evaluations, giving the average cost, time, and output tokens to complete a single Intelligence Index task

3. Cached input token reporting:
We now report cached input tokens and their impact on cost, including the cost to run the Intelligence Index, to better reflect the real cost of running each model

Key Results:
➤ Leading models: Claude Fable 5 (with Opus 4.8 fallback, 60) leads the Artificial Analysis Intelligence Index v4.1 by four points but is currently unavailable, leaving Claude Opus 4.8 (max, 56) as the most intelligent available model, ahead of GPT-5.5 (xhigh, 55) ➤ Open weights leading models: Among open weights models, DeepSeek V4 Pro (max, 44) and MiniMax M3 (44) lead, followed by Kimi K2.6 (43) and MiMo-V2.5-Pro (42)
➤Cost per Task: Claude Opus 4.8 (max) is the most expensive available model at $1.78 per task, with Claude Fable 5 the highest overall at $3.25. GPT-5.5 (xhigh) scores within a point of Opus 4.8 on the Intelligence Index at $0.99 per task. DeepSeek V4 Pro (max) stands out on the Intelligence vs Cost per Task chart at $0.04 per task, with other leading proprietary models costing 20x to 45x more
➤Time per Task: time per task (inference decode time) ranges from 1.5 minutes for Grok 4.3 (high) to 13.5 for Claude Sonnet 4.6 (max), a roughly 9x spread. Claude Opus 4.8 (max) completes a task in 6.4 minutes and GPT-5.5 (xhigh) in 3.7, while Gemini 3.1 Pro Preview stands out on the Intelligence vs Time per Task chart at 1.6 minutes for a score of 46

[Aric Mitchell @AricMitchell]

I've come to realize that we tend to see AI as a genius in our weak areas and a fraud in our strong ones.

[Aric Mitchell @AricMitchell]

OK, that's hilarious. And felt.

[Ken Davis @KenDavi66334773]

@ajitcodes I just pasted my salary into Claude and it told me to kill myself

[Aric Mitchell @AricMitchell]

Cool new feature from @GeminiApp

[Google Gemini @GeminiApp]

You can now create and edit images directly in Gemini Live.

Whether testing out room decor, getting help with math, or creating shareable memes, it all happens in real-time.

Just open the Gemini app, tap the Live button, share your camera, and tell Gemini what you want to see.

[Aric Mitchell @AricMitchell]

I know I'm probably just stating the obvious here, but release day is usually not the best time to try a tool like @Google Spark. Been buggy AF all day. Check back in two weeks.

[Aric Mitchell @AricMitchell]

I get the sentiment behind "Preferred Sources," @GeminiApp, but it seems like just another way to build your own echo chamber, this time with AI. https://t.co/1iPm75UkMl

[Aric Mitchell @AricMitchell]

@OpenAIDevs Very frustrated I'm not getting these cool adds as a Windows user.

[Aric Mitchell @AricMitchell]

Some of these are absolutely hilarious.

[Não Intendo @blognaointendo]

Tem uns aí que eu não lembrava

[Aric Mitchell @AricMitchell]

We have AIs that can plan your day, run your business, solve difficult mathematical equations, and write entire books with minimal input from you. Yet, here I am, more amazed by a mouse pointer.

[Google DeepMind @GoogleDeepMind]

We’re reimagining a 50-year-old interface - the mouse pointer - with AI. 🖱️

These experimental demos show how people can intuitively direct Gemini on their screens using motion, speech, and natural shorthand to get things done 🧵

[Aric Mitchell @AricMitchell]

@KieranGilmurray I’d probably add a fifth board-level question to the four listed here: “Do the people overseeing this technology actually use it themselves, and, if so, what is the most complex thing they can do with it?”

[Aric Mitchell @AricMitchell]

The best ChatGPT use case for most office workers is still boring:

Paste messy notes.

Ask for decisions, open questions, follow-ups, risks, and a short recap email.

That one workflow beats 90% of the “AI will change everything” noise.

[Aric Mitchell @AricMitchell]

"Act like an expert" is a weaker prompt than it sounds.
Try this instead: "List what a careful expert would verify before answering. Then answer only after separating known facts from assumptions." #grok #chatgpt #prompting

[Aric Mitchell @AricMitchell]

This 💯
Never used OpenClaw, but it inspired me to ask ChatGPT/Codex to handle more challenging, seemingly "impossible" things. Pretty much whatever I don't want to do, "Why don't you just do it for me?" A lot of times, it works. Either way, it teaches me a bunch along the way.

[Edison Ade @buzzedison]

Peter Steinberger created OpenClaw, an open-source agentic AI assistant. He released it publicly, it blew up, Sam Altman saw it, and now he works at OpenAI. That all happened in less than a year. The real change is not AI writing your articles but handling complex, multi-step tasks with smart decision-making: digging up info, picking the best route, executing, and sending a final report.

The productivity gains from single prompts are rounding errors compared to what persistent agents will produce.

[Aric Mitchell @AricMitchell]

@DanielMiessler well said, sir