The launch vibe for GPT-6.1 Terra was upbeat about speed and price. People who used it for computer use, Blender work and everyday tasks described it as fast and cheap, often next to GPT-6.1 Sol or Claude Sonnet 5.5. The doubts came from coding and longer jobs, with reports of weaker results below maximum effort, tasks stalling in Codex, and Claude models doing better on some head-to-head builds. Of the 22 posts quoted here, 10 are positive, 7 mixed, 4 negative and 1 neutral.
Peter Gostev@petergostevPositive
“Feels like a slightly simpler 6.1-Sol, which is honestly pretty good.”
Antikythera@Antikythera_0Mixed
“not a model for coding, but a model for working”
Flavio Adamo@flavioAdPositive
“I've really come to appreciate how it writes! (and it's fast)”
Zach@ZryMillerPositive
“First impressions are it is extremely fast and very cheap.”
Salio@Mr_SalioNegative
“Terra is one of the most disappointing models I've tested since GPT-5.5 … This model only has a crazy speed”
adi@adidoitPositive
“underrated release - smart, fast, cheap (used very little of my quota) … seems to be a great subagent for 6.1 sol”
Matt Workman@mattworkmanPositive
“GPT 6.1 Terra (Ultra) is INSANELY good at 3D Blender computer use”
“This thing absolutely flies through a computer. … Probably the best computer-use model I’ve tested.”
Wayne Lowry@wikiwayneNeutral
“7 of 7 loaded and ran on the first try, 41 cents total, about 6.8 minutes. … Pass means it ran, not that it's good.”
Paweł Huryn@PawelHurynMixed
“It's cost and time effective. But right after max, its performance drops dramatically - far below Haiku 5.5 (xhigh).”
bimi Ⓥ⏩⏭️∞@DrBimiNegative
“I tested Terra and found it to be absolute garbage for my use case. It just invents new rules instead of following the ones already in the code”
Naomi Metzger@naometzgerMixed
“it did a good rough cut, but i wanted more control”
NinjaCodex@N1njaCodexMixed
“Opus delivered as always … But Terra did a great job for the price too - a cozy winter world for almost 6x less”
WRONG ADVICES@Helios_RisesMixed
“it actually provides great value as a worker agent (90% of Opus 5.5's solo score at 3% of the cost and 9% of the time), but is much weaker as an orchestrator or solo operator.”
Serafim Batzoglou@s_batzoglouPositive
“6.1 Terra is at around Claude Sonnet 5.5 level … At $5/M tokens, this is fantastic.”
Pranav Reddy@AI_ScreeningNegative
“haiku feels better than 6.1 Terra”
Trí ACJ 🇻🇳@TriACJ1409Mixed
“Our first impression: it’s really fast. … It seems to use up a lot of my 5-hour quota, even though the token usage is much lower than with other models”
neo yang@NYang40764Positive
“Faster and cheaper than GPT 6.1 Sol, with basically the same quality.”
Vinh Nguyen@vinhnxPositive
“GPT-6.1 Terra has been impressive at computer use.”
Petra@synthshareaiNegative
“GPT Terra 6.1 is unusable in Codex. Stops tasks randomly.”
Sahil Panhotra | Indie Builder@SahilPanhotraMixed
“this thing was flying yesterday when I tested it. now it's already losing that speed”
