LOADING DATASHEET
LOADING DATASHEET
by OpenAI
RANKED #42 OF 316 ASSISTANTS · OVERALL #59 OF 493 RANKED · VIBE SCORE 6.4 · 17,359 VOICES ·
COPY BADGE pastes a README snippet with the rank SVG.
A fast, budget-friendly AI helper designed for writing and coding tasks.
829 mentions
885 mentions
26 weeks · 17,740 voices in this window
HOLDING UP
Recent vibe is steady. No clear fade in the crowd.
Closed model. OpenAI hosts it, so you use it through an app or an API.
GPT 6 Luna is closed, so you rent it. These open-weight models can genuinely stand in for it. Download once, own forever.
Users love GPT 6 Luna for its fast responses, lower pricing, and strong writing and coding skills. However, some feel it is a downgrade from older versions, and complain about outages, occasional hallucinations, and preachy refusals.
6536 POSTS · 10823 COMMENTS · REDDIT · GITHUB · HACKER NEWS · DEV FORUMS · DEV.TO · BLUESKY · BLOG · LEMMY
31 thumbs up · 16 thumbs down
27 thumbs up · 15 thumbs down
1 thumbs up · 35 thumbs down
4 thumbs up · 24 thumbs down
6 thumbs up · 19 thumbs down
7 thumbs up · 2 thumbs down
4 thumbs up · 1 thumbs down
4 thumbs up · 0 thumbs down
“GPT 6 Luna is a significant downgrade from gpt 5.6.”
Checked picks first: tags say why you would switch, stars say how fully each one stands in for GPT 6 Luna.
That didn't backfire at all
Anthropic CEO Dario Amodei: AI Biology Will Go From Weak To Superhuman Within A Few Years
Luna 6 is a massive downgrade over Luna 5.6. Misses crucial details, less helpful than 5.6 Luna
GPT 6 Luna Max scores as high as GPT 6 Sol High and Astra Low whilst being 3x cheaper than GPT 6 Sol and 7.3x cheaper than Astra on DeepSWE
Anthropic needs an even cheaper model than Haiku
GPT6 Luna in my codex already
OpenAI launches GPT-6 Sol + Luna: Astra intelligence moves down the cost curve
One Shot Wolfenstein 3D - Opus 4.6 vs. Opus 5.5
I want Mistral to succeed. But we need to talk about pricing, models, reasoning, Vibe, and communication.
The harness is absolute trash
Codex Rate Limits Discussion Thread
Codex Rate Limits Discussion Thread