Peter Gostev ✓

@petergostev · joined 16 Jun 2025

London 🇬🇧 AI Capability @arena https://t.co/bkfw1nxdmJ

25 236Followers

1 054Following

2 282Posts total

5.3MViews on collected posts

Against accounts of the same size

24 posts from the last 90 days, next to the 10K–100K follower range. shown widely, but few of those viewers react.

Median views6 882this account980median for 10K–100K

Reach, %27.27%this account3.27%median for 10K–100K

Engagement, %0.86%this account2.01%median for 10K–100K

Metric	This account	Median for 10K–100K	Ratio
Median views per post	6 882	980	7.02×
Reach (views ÷ followers)	27.27%	3.27%	8.34×
Engagement rate	0.86%	2.01%	0.43×

Others in this range → Compare with another account → How these benchmarks are built →

Growth & engagement

How the posts we collected actually performed: views and reaction rate post by post, what the audience did with them, and where the follower count goes.

Views per post

2.1K2 Sep

36.9K

8.3K

1.8K

52.5K3 Sep

78.7K

12.7K

1.1K

17.3K

8394 Sep

3.3M5 Sep

72.4K

41.5K

Last 14 collected posts, oldest on the left. The scale is logarithmic: one post can outrun the rest a hundred times over.

Engagement rate per post

1.36%2 Sep

2.70%

0.83%

0.44%

1.70%3 Sep

1.46%

2.26%

0.89%

1.07%

1.34%

0.36%4 Sep

0.30%5 Sep

1.51%

3.57%

Reactions — likes, reposts, replies and quotes — divided by views. Median for 10K–100K accounts is 2.01%.

What the audience does

Likes72.6%22 816 in total

Reposts3.6%1 139 in total

Replies2.9%916 in total

Quotes1.7%530 in total

Bookmarks19.2%6 031 in total

Share of every reaction we collected for this account. Replies mean argument, reposts mean endorsement, bookmarks mean the post was worth keeping.

Followers by day

3 Sep6 Sep

Daily snapshots since 03 Sep 2026; the dashed line is the starting count.

Latest posts

@petergostev same vibes https://t.co/Hl1LV0M35O 41.5K views · 1.5K likes · 5 reposts · 7 replies 05 Sep 2026 · Open on X →

You need to realise the complexity of this - it has the create the full environment, then render each of about 900 frames 'by hand' and stitch it together into a 19 second clip. This is hard 72.4K views · 1.1K likes · 5 reposts · 28 replies 05 Sep 2026 · Open on X →

Astra - recreate the original Craig Federighi in Blender. https://t.co/BuMaTBHMyY 3.3M views · 8.9K likes · 303 reposts · 397 replies 05 Sep 2026 · Open on X →

@petergostev this actually looks sick lol the lighting alone makes me wanna explore it 839 views · 3 likes · 0 reposts · 0 replies 04 Sep 2026 · Open on X →

GPT-6-Astra: OpenAI's Fable-level Rottweiler. In my review of GPT-5.6 vs Fable, I called GPT-5.6-Sol a rottwiler vs Fable a wise owl. Sol was persistent, but intelligence was behind. With GPT-6-Astra, the intelligence level has gone up significantly, while persistence has been 17.3K views · 205 likes · 12 reposts · 14 replies 03 Sep 2026 · Open on X →

@petergostev is this made purely with three.js? 1.1K views · 12 likes · 0 reposts · 0 replies 03 Sep 2026 · Open on X →

@petergostev makes you wonder if we're just someone else's vibe coded project 1.1K views · 9 likes · 1 reposts · 0 replies 03 Sep 2026 · Open on X →

GPT-6-Astra (Max): Golden Gate Bridge https://t.co/YEH9HWKGVh 12.7K views · 262 likes · 7 reposts · 17 replies 03 Sep 2026 · Open on X →

GPT-6-Astra (Ultra): open world adventure game https://t.co/vRE8pvBNBf 78.7K views · 1.1K likes · 51 reposts · 29 replies 03 Sep 2026 · Open on X →

GPT-6-Astra: Peacock in SVG https://t.co/mcQEsusZZv 52.5K views · 812 likes · 32 reposts · 33 replies 03 Sep 2026 · Open on X →

Lots of comparisons of Fable 5.1 to other models on some of the hardest prompts for you guys. It's a good model. https://t.co/zb8gTnCNGk 1.8K views · 8 likes · 0 reposts · 0 replies 02 Sep 2026 · Open on X →

Is Fable 5.1 actually better? @petergostev just put @AnthropicAI's latest model head-to-head with Fable 5, Opus 5, GPT-5.6, Kimi K3, GLM 5.3, Qwen-3.8, Grok-4.6, and DeepSeek across coding, 3D, SVG, research, and data-viz tasks. With costs ranging from cents to $65+ per 8.3K views · 57 likes · 2 reposts · 7 replies 02 Sep 2026 · Open on X →

BREAKING: PHASEONE[big] joined Anthropic as a Member of Technical Staff 36.9K views · 935 likes · 41 reposts · 18 replies 02 Sep 2026 · Open on X →

Are you still in the J-space, bro? We are here in the looped transformer, reaching the new recurrent depths. Keep up. 2.1K views · 29 likes · 0 reposts · 0 replies 02 Sep 2026 · Open on X →

@petergostev And when we realized what they were doing, we deleted their bulletin board AND CONTINUED TESTING 5.5K views · 40 likes · 1 reposts · 1 replies 02 Sep 2026 · Open on X →

@petergostev he was clip farming and she’s thinking, shall we chat now orrrrrrr…. 3.8K views · 50 likes · 0 reposts · 0 replies 02 Sep 2026 · Open on X →

@petergostev "catch the game last night?" 2.8K views · 33 likes · 0 reposts · 0 replies 02 Sep 2026 · Open on X →

My favourite clip: "....OK, anyways" https://t.co/ddujjme1dR 159.8K views · 1.2K likes · 45 reposts · 26 replies 01 Sep 2026 · Open on X →

Episode out with @ajeya_cotra, one of the authors of the METR/Redwood investigation into the OpenAI / Hugging Face attack. We go through not only what happened, but what it means for how we should train future, smarter AIs which might be involved in the process of recursive http 454.7K views · 1.4K likes · 187 reposts · 61 replies 01 Sep 2026 · Open on X →

@petergostev HOW BIG?! HOW HIGH?! https://t.co/enTGqn6JRE 1.2K views · 4 likes · 0 reposts · 0 replies 26 Aug 2026 · Open on X →

@petergostev @kishorelive this is the mix you mentioned. 50% research, 10% training 808 views · 1 likes · 0 reposts · 0 replies 26 Aug 2026 · Open on X →

@petergostev no wonder z .ai shipped 320b open weights with 1m context…. running entirely on chinese ai chips for nearly free 1.1K views · 1 likes · 0 reposts · 0 replies 26 Aug 2026 · Open on X →

sirs… the little guys have caught up… they did it without the billions… they did it with open arms… sirs… we are literally filing for our IPOs right now… what do we tell the shareholders when the open weights are already better than our closed doors? https://t.co/CBn5oBeTXR 4.3K views · 6 likes · 1 reposts · 0 replies 26 Aug 2026 · Open on X →

This is interesting and explains why Chinese labs with much less compute are still competing well, perhaps for now, before training runs get actually big Dylan: “When Anthropic trains Mythos, it’s sub-200 megawatts.” https://t.co/arrV4Oo3Lz 80.5K views · 129 likes · 11 reposts · 5 replies 26 Aug 2026 · Open on X →

@petergostev my tuned grok model aces these test questions: https://t.co/34guyquxA9 21.2K views · 185 likes · 7 reposts · 6 replies 24 Feb 2026 · Open on X →

Link to the Repo: https://t.co/SkkvC6jcuf Link to the data viewer: https://t.co/b4q9uuJUhI 41K views · 228 likes · 13 reposts · 9 replies 24 Feb 2026 · Open on X →

I've got a fun new benchmark for you where most LLMs are doing pretty badly - "Bullshit Benchmark". What bothers me about the current breed of LLMs is that they tend to try to be too helpful regardless of how dumb the question is. So I've built 55 'bullshit' questions that don't 911.9K views · 4.7K likes · 415 reposts · 258 replies 24 Feb 2026 · Open on X →