modelsentiment
← Leaderboard

GPT-4

OpenAI · GPT / ChatGPT series · released 28 May 2023 · as of 13 Nov, 20:12 UTC

No index yet

6 opinions name GPT-4 last 7 days; an index needs 10. 52 posts and comments mention it. Until then the page shows its mentions, benchmarks and threads, and where the 44 rated models stand.

0255075100

Per day

The last 14 UTC days. A day shows an index from 10 opinions.

  1. Sat 31 Oct0
    0 opinions, no index below 10
  2. Sun 1 Nov0
    0 opinions, no index below 10
  3. Mon 2 Nov2
    2 opinions, no index below 10
  4. Tue 3 Nov1
    1 opinion, no index below 10
  5. Wed 4 Nov1
    1 opinion, no index below 10
  6. Thu 5 Nov0
    0 opinions, no index below 10
  7. Fri 6 Nov0
    0 opinions, no index below 10
  8. Sat 7 Nov0
    0 opinions, no index below 10
  9. Sun 8 Nov1
    1 opinion, no index below 10
  10. Mon 9 Nov1
    1 opinion, no index below 10
  11. Tue 10 Nov0
    0 opinions, no index below 10
  12. Wed 11 Nov2
    2 opinions, no index below 10
  13. Thu 12 Nov1
    1 opinion, no index below 10
  14. Fri 13 Nov1 so far
    1 opinion, no index below 10

Is GPT-4 getting worse?

Too few opinions to say. A full read needs 30 opinions from 20 threads and 20 accounts; GPT-4 has 6 opinions from 6 threads and 6 accounts last 7 days.

Benchmarks

Its rank on each list, bottom to top; no Reddit row until it has an index.

  1. #125 of 128 on Arena
  2. #100 of 107 at Artificial Analysis
  3. #67 of 69 at Artificial Analysis

$30 in / $60 out per million tokens · 8k context · no reasoning mode · OpenRouter, 13 Nov

Per subreddit

The model's index within each subreddit, from 10 opinions; faded below 30 opinions, 20 threads or 20 accounts.

Threads behind the opinions

The threads with opinions about this model. A thread's index reads the opinions in it, not the post, and shows from 10 opinions.

ThreadSubredditOpinionsIndex
I lost my friend last night: the new ChatGPT changes are for realin 1 commentr/ChatGPT1–
After four years of development, are LLMs capable of writing high-quality prose?in 1 commentr/SillyTavernAI1–
In the spirit of understanding the "opposition"in 1 commentr/singularity1–
The hall of fame as of november 2026in 1 commentr/singularity1–
METR's Ajeya Cotra: "This is the most important update we have gotten about AI risk in the last couple of years — much bigger than the Hugging Face attack." ... "implies extreme superintelligence within a year."in 1 commentr/ChatGPT1–
No GPT-6.2, no Fable 5.5, no Qwen 4in 1 commentr/singularity1–

Weekly email

Every Monday: the warmest and coldest models, and who moved beyond noise. No tracking; one click unsubscribes.