The ANCEL is the pick. At $32 it was the least expensive tester to pass or top every graded category, and it was the most repeatable of the group: disconnect it, reconnect it, and it returned 810 CCA, 12.87 V, 3.61 mΩ, 95% state of health, and 100% state of charge a second time without a single digit moving. It read voltage within 0.03 V, correctly condemned the worn-out battery at 22% health, passed the in-vehicle battery, cranking, and charging tests, and asked for battery temperature before judging the battery chilled to 0°F.
Three other testers earned a recommendation. The AUTOPHIX at $48 matched the ANCEL on accuracy and adds an alternator test with copper-jawed clamps. The TOPDON BT100 at $59 performed just as well and has by far the most robust clamps here, though that is roughly twice the ANCEL's price for no measurable gain. The FOXWELL BT705 at $117 read 12.90 V dead on, got every diagnosis right, and reports cranking voltage and time, alternator output, and no-load and loaded voltage — the most complete picture on offer, at a price to match.
Spending more is no guarantee. The $90 TOPDON BT200, sold on better software than the BT100, called for replacement on a healthy battery that was merely cold, and so did the $300 Midtronics, which also sets cranking amps only in 20-amp increments and leaves the cranking and charging procedures to the manual. At the other end, the OBD Resource and KINGBOLEN read the new battery at 740 and 749 CCA against an 830 CCA rating and both missed state of charge on a fully charged battery, the BT300 was 0.22 V off and wrongly failed a good AGM battery in the vehicle, and the V-PEAK returned the same incorrect result four times in a row on that same test. The MOTOPOWER 514A is accurate for what it is — voltage, and nothing else — so it is a $13 voltmeter rather than a battery tester. If the goal is knowing whether a battery is genuinely bad before paying for a new one, $32 covers it.