At what voltage AA battery needs binned?

[Kleinstein]

For alkaline batteries measuring the voltage is good.

If you still run across an old style dry cell (rarely a good choice anymore, but they sometimes still come with cheap stuff) measuring the voltage with no load does not help. An old used up dry cell still shows ~ 1.5 V open circuit, but the internal resistance goes up. So one needs to test those old cell with some load (e.g. 1 K).

The same applies to CR2032 and similar Li primary cells - even when empty they still show some 3.1 V open circuit, but the voltage drops with load. When really old even the 10 M from a DMM can make a difference.

[wraper]

The same applies to CR2032 and similar Li primary cells - even when empty they still show some 3.1 V open circuit, but the voltage drops with load. When really old even the 10 M from a DMM can make a difference.

Those are not empty, they are bad for other reasons. Like bad internal contact or whatever causes very high internal resistance. Measuring ESR at AC would be better than loading them.

[Psi]

I tend to just short non-rechargeable AAs on the 10A/20A DMM range and see what the amps is (Only a quick tap, like for 250ms).
If i see above 3A then I consider it ok. If it's below 1A it goes in the trash. 
I find this tells you more than voltage. It's more of an ESR test. But some people don't like shorting them like that.

Shorting a rechargeable AA will easy push 10A so probably not a good idea, and a non-rechargeable lithium alkaline is probably not a good idea either if your DMM fuses are expensive    I've never blown a 10A DMM fuse doing this but YMMV.

I usually do a voltage test first when I think it could be a fresh cell (1.5V etc) But if it's <=1.3V  I do the 10A scale shorting test to see what it's current sourcing ability is.  If I know it's not a fresh cell then i go straight to the short test.

I would never do a short test like that with LiPo/LIon/NiCa cell.   NiMh is borderline
If i wanted to check current sourcing on those i would insert a load resistor

[ipman]

Just my humble opinion: while I support the idea of a load, a dead short is probably not the best approach.
Overall there could be some of them with high internal resistance which could hopefully be moved to less demanding duties like remotes, clocks and they could continue there for quite a while. If they don't leak in a remote there is no danger to notice they are out of juice e.g.

[Psi]

Yeah, I'm not suggesting others do it, just saying what i do 

Remotes can take a fair bit of current when transmitting due to using high current IR leds. They just do it so rarely and for such a short time. A good remote will have enough caps to store that charge, but they often don't.

If you can't get 1A out of it from a dead short then the cell is pretty bad.

[electric_cheeze]

I tend to just short non-rechargeable AAs on the 10A/20A DMM range and see what the amps is (Only a quick tap, like for 250ms).
If i see above 3A then I consider it ok. If it's below 1A it goes in the trash. 
I find this tells you more than voltage. It's more of an ESR test. But some people don't like shorting them like that.

Shorting a rechargeable AA will easy push 10A so probably not a good idea, and a non-rechargeable lithium alkaline is probably not a good idea either if your DMM fuses are expensive    I've never blown a 10A DMM fuse doing this but YMMV.

I usually do a voltage test first when I think it could be a fresh cell (1.5V etc) But if it's <=1.3V  I do the 10A scale shorting test to see what it's current sourcing ability is.  If I know it's not a fresh cell then i go straight to the short test.

I would never do a short test like that with LiPo/LIon/NiCa cell.   NiMh is borderline
If i wanted to check current sourcing on those i would insert a load resistor

I use a cermaic 1Ω 5W resistor and a voltmeter;
Step 1 measure battery Open-Circuit Voltage (Voc​)
Step 2 Measure Voltage loaded (Vld​) with 1 ohm load.
Internal resistance Ri=
Ri​= RΩ​ X (Voc - Vld / Vld)
or Ri = 1 X (Voc - Vld / Vld)
  or Voc - Vld / Vld

[CatalinaWOW]

I don't have formal data to support this, but my experience with alkaline cells has lead to this way to categorize them. 

Voltage above 1.4  -  Probability of leak is higher than I would like, but I can live with it.
Voltage    1.2-1.4   -  Most of my battery powered thingies will work, but leak probability rules out using them in good things.
Voltage  below 1.2  -  Not enough things will work long enough to be worth using, and leak probability is very high.  Into the bin.

The best if used by date is a joke.  I have had a large unopened package of major brand AAs show severe leakage in more than half of the cells several years before the expiration date on the package. 

[Jeroen3]

You should get a batteriser 

[default0.0player]

Unlike rechargeables, disposable batteries typically have higher and higher internal resistance as the battery level goes down.
At higher loads, the difference between open-circuit and loaded voltages are larger, such as 1.5-1.4V near full but 1.3-0.9V near empty.

source: https://lygte-info.dk/info/batteryEnergyAtLowVoltage%20UK.html
The voltages listed on the table are loaded. For example if you measured 1.3V on a voltage meter, the battery is only providing 1.1V for a 1W device