noborus/trdsql | ProgramBench

CLI tool that can execute SQL queries on CSV, LTSV, JSON, YAML and TBLN. Can output to various formats.

1,312

Generated Behavioral Tests

97.4%

Best Score

Claude Opus 5.5 (xhigh)

Results by Model

Hover for details · Click a model to open it · Drag to zoom the x-axis · Shift-drag to pan · The line marks the Pareto frontier (best score per cost)

30 runs

#	Model	Score	Cost	Calls
1	Claude Opus 5.5 (xhigh)	97.4%	$20.79	223	trace →
2	Muse Wonder Full	96.0%	TBA	475	trace →
3	Muse Wonder (xhigh)	95.7%	$30.87	362	trace →
4	Claude Fable 5.1 (xhigh)	94.8%	$33.87	220	trace →
5	GPT 6 Astra (max)	93.3%	$8.18	42	trace →
6	Claude Opus 5 (xhigh)	92.5%	$50.60	307	trace →
7	Claude Opus 4.8 (xhigh)	84.1%	$15.68	136	trace →
8	GPT-5.6 Sol (xhigh)	81.4%	$4.78	39	trace →
9	Muse Spark 1.3 (xhigh)	79.5%	$2.58	341	trace →
10	Gemini 3.7 Flash	78.0%	$1.58	179	trace →
11	GLM-5.2	76.9%	$30.57	214	trace →
12	GPT 5.5 (xhigh)	76.4%	$7.07	70	trace →
13	Muse Spark 1.3 (max)	75.4%	$9.09	349	trace →
14	Gemini 3.6 Flash	73.6%	$3.64	162	trace →
15	Gemini 3.5 Flash	73.0%	$3.64	147	trace →
16	Muse Spark 1.1 (xhigh)	72.6%	$0.56	146	trace →
17	Muse Spark 1.2 (xhigh)	72.3%	$1.00	171	trace →
18	GPT-5.6 Sol	71.3%	$0.84	15	trace →
19	GPT 5.5 (high)	71.0%	$2.11	22	trace →
20	Claude Opus 4.6	66.8%	$9.89	307	trace →
21	Claude Opus 4.7	61.7%	$3.13	73	trace →
22	GPT 5.5	54.0%	$0.74	13	trace →
23	GPT 5.4	49.1%	$0.38	14	trace →
24	Gemini 3.1 Pro	46.4%	$1.10	96	trace →
25	Claude Haiku 4.5	25.2%	$1.70	207	trace →
26	Gemini 3 Flash	14.9%	$0.26	109	trace →
27	Claude Opus 4.7 (xhigh)	3.4%	$8.71	175	trace →
28	GPT 5.4 mini	3.4%	$0.03	12	trace →
29	Claude Sonnet 4.6	3.0%	$12.60	413	trace →
30	GPT 5 mini	2.9%	$0.01	14	trace →

Click a row to replay how that model rebuilt this program, or the model name to open its full run