Average speeds at Q4 quantization. Actual performance varies by model architecture and context length.
3B
—
1.7GB needed
7B
—
3.9GB needed
14B
—
7.9GB needed
32B
—
18.0GB needed
70B
—
39.4GB needed
Measured rig reports
We rent the machine and time every model on it: decode, VRAM peak, concurrency, watts and cost per million tokens. Yours may already be one of them — and two are free to read in full.
Neighbours in memory rather than in price: memory decides whether a card can do the job at all, so two cards of the same size at different prices is the comparison you are making. Ordered by memory, then bandwidth — not by the value column, which is worked out from bandwidth and price alone and therefore rewards a cheap card whatever its software stack does to that bandwidth in practice. Read it as one input, not as a ranking.
Device under test
Apple A14 GPU — 4 GB unified, 1 GB usable.
Apple A14 GPU spec
Brand
Apple
Unified memory
4 GB
Usable by a model
1 GB
Bandwidth
25.6 GB/s
FP16 Compute
1 TFLOPS
TDP
5 W
Architecture
Apple GPU 4-core
AI capability
4/ 498 models @ Q4
With 1 GB of its 4 GB reaching a model and 25.6 GB/s bandwidth, this machine handles models up to 0.14B parameters.
Speed ≈ bandwidth / model_size × efficiency. A 7B model at Q4 runs at ~3 tok/s.