The only free+great+.com lane
The distant-leap blend
Two real words from far-apart domains (the ByteDance move). Otherwise: BUY a two-word brandable, or DROP-CATCH a great expiring .com.
Market research
Market trends
The headline, stated plainly
The free .com pool for good names is ~EXHAUSTED. ~750 candidates checked 2026-07-03..05: ALL real dictionary words, ALL first-order A1 compounds (gold/top/star/king/echo/spark × word/talk/rank/say), ALL iconic phrases (namedrop, pageone, lastword, hearsay), ALL Latin single words = registered or parked ($20k-$39M asks). Re-confirmed 2026-07-06 with ~2,250 MORE .com checks: every name that CLICKS is taken; every FREE one is mush (prowroar/keelleap/answermade). The free-real-word-.com door does NOT exist.
The facts underneath
Verified against the live registry, cited to their sources, updated whenever re-checked.
Speed
Open seams close in DAYS
drumgold + bravenzo were registered days before the hunt found their lanes. Verified-available = register same-day or lose it.
The construction quadrant
18/26 free on .ai
Lexically-empty + phonologically-dense coinages (round 13) vs 24/25 TAKEN for real words — the domain wall is gone in this quadrant.
Coined vs real words
2x trademarkability · +31% 48h recall
BrandScout 2026-03, 4,000 brands: coined names following natural phonotactics outperform real-word names.
Staleness
Every availability list is a snapshot
Re-check at the registry before acting.
The zeitgeist, 2023–26
What actually won in the market — and what the winners refused to look like.
Winners
- Real English words with meaning
- Serious composites
- Iconic short oddballs
- First names
- Animals
Anti-patterns
- -ly
- -ify
- misspellings
- zero-anchor -a/-ia mush
Source: Crunchbase era map 2025-05 (via the pipeline-v2 zeitgeist sweep).
Why LLM names fail — the averaging diagnosis
The problem
LLM-as-generator = LINGUISTIC AVERAGING. Models converge on the statistical centre of "what a good name looks like" and produce consensus mush: Flux / Pulse / Axera.
The weighting
Method ≈ 70% · brief ≈ 20% · model ≈ 10% of the failure.
The fix
AI is a PROFILER, never the generator. Generate from morphemes + phonesthemes with a REQUIRED one-line morpheme story per candidate; the engine generates, the panel discriminates.
Sources
- Tanj 2026-03 (linguistic averaging)
- Focus Lab × Placek 2026-01 (Identify–Invent–Implement; AI-as-profiler; 107 linguists + 7 in-house LMs)
- NameStormers 2026-01 (name-slop filter)
- River+Wolf 2024/26 (prompt tonality, lateral paths, VC-portfolio zeitgeist mining)
- Klink JCR + JCPY 2024 (phonestheme evidence)
- Vitevitch & Donoso 2012, PMC3289729 (phonotactic probability raises purchase intent)