Newick: coffee cultivar tree, by gpt-6-luna
gpt-6-luna · round 1 · seed 3$0.00SVGtree.nwk
tree.nwk · 2 lines
((Bourbon:0.75,Typica:0.75,Caturra:0.75,Catuai:0.75):0.25,(((Mundo_Novo:0.1875,Maragogipe:0.1875):0.1875,(Pacas:0.1875,Villa_Sarchi:0.1875):0.1875):0.375,((SL28:0.1875,SL34:0.1875):0.1875,(Gesha:0.1875,Java:0.1875):0.1875):0.375):0.25);

Bourbon, Typica, Caturra and Catuai level in the split of four, though some come from others there.

modelopenai/gpt-6-lunatasknewick-coffee-cultivars@v1, checks c1, harness h1date2026-10-14rounds1 of 8 allowed, 12 minutes on the clockcost$0.0007thinkingmediumtokens6,363 in, 1,147 out, of which 883 reasoningtime20 s
Valid Newickdecided by a number, from the filepass
Twelve cultivarsdecided by a number, from the filepass
Every tip leveldecided by a number, from the filepass
One split of fourdecided by a number, from the filepass
read the post
its other runs
the prompt it was given
Write an ultrametric Newick tree of twelve coffee cultivars with exactly one polytomy, of four.
Frozen at v1. This model got this and nothing else.
ground truth

The tree read with Biopython, every path from the root to a tip added up from its branch lengths, and the children of every node counted. The checks only read the brackets. Whether a cultivar sits beside its relatives is not checked here, and the post of 14 October looks at that by eye. How it is scored, and this run's files on GitHub.