collinear-ai/yc-bench
1
0
Fork 0
mirror of https://github.com/collinear-ai/yc-bench.git synced 2026-04-28 17:29:35 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Added the configs and updated the results.

Browse source
This commit is contained in:
adit jain 2026-02-26 13:37:58 -08:00
parent 5d2962073d
commit 3643806dce
22 changed files with 14168 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

11
scripts/plot_comparison.py
View file

@ -1,4 +1,4 @@
"""Sonnet 4.6 vs Gemini 3 Flash — apples-to-apples comparison plot."""
"""Sonnet 4.6 vs Gemini 3 Flash vs GPT-5.2 — apples-to-apples comparison plot."""
import sqlite3
from pathlib import Path
from datetime import datetime
@ -17,13 +17,16 @@ MODELS = {
"slug": "anthropic_claude-sonnet-4-6",
"label": "Sonnet 4.6",
"color": "#2563eb",
"dash": "-",
},
"gemini": {
"slug": "gemini_gemini-3-flash-preview",
"label": "Gemini 3 Flash",
"color": "#f97316",
"dash": "-",
},
"gpt52": {
"slug": "openai_gpt-5.2",
"label": "GPT-5.2",
"color": "#16a34a",
},
}
@ -89,7 +92,7 @@ def load_all():
def make_plot(runs):
fig, axes = plt.subplots(3, 3, figsize=(18, 14), facecolor="white")
fig.suptitle(
"Sonnet 4.6 vs Gemini 3 Flash · YC-Bench · 1-Year Horizon",
"Sonnet 4.6 vs Gemini 3 Flash vs GPT-5.2 · YC-Bench · 1-Year Horizon",
fontsize=16, fontweight="600", y=0.98, color="#1a1a1a",
)