modelsentiment
← Leaderboard

GPT-5.6 Terra

OpenAI · GPT Terra series · released 9 Jul 2026

Sentiment · since 25 Sep
47
no clear lean
from 15 opinions
Mentions
70
posts and comments

Sentiment per day

Mentions per day

Reddit vs benchmarks

  • #31 of 40
  • #33 of 98
  • #32 of 76
  • #14 of 49
  • #6 of 57
  • #12 of 44
  • #6 of 74

$2 in / $12 out per million tokens · 1.05M context · reasoning mode

Per subreddit

Threads behind the opinions

ThreadOpinionsIndex
Rate my stack2–
GPT 6 sol isnt as good anymore?2–
I built a classifier tool for Codex, claude-code, Pi and OpenCode after finding out one triage cost me 100k tokens1–
Qwen3-VL 8B on a MacBook vs Opus 5.5 / Sonnet 5 / GPT-5.6 on 137 messy documents: beat GPT-5.6 on tax forms, lost badly on Indian date formats1–
Qwen3-VL 8B on a laptop vs Opus 5.5 / Sonnet 5 / GPT-5.6 on 137 messy documents: beat GPT-5.6 on tax forms, lost badly on Indian date formats[R]1–
Codex is in a really bad spot right now. Opus 5.5 > all OpenAI's models for coding, and much more generous rate limits1–
GPT-6 Sol is massive downgrade1–
Luna 6 isn't as bad as the whiners say1–
Well that was quick...1–
Do we get reset for every 1 million Users Codex loses to Anthropic?1–
After using the Sol-6 for a few days, the verdict is: it’s a Terra 5.6 with post-tuning1–
Impressed with Opus 5.5 coming from Codex1–
The problem with Sol 6 isn’t the price. It’s the wasted time.1–