Meta MTIA 200 vs Meta MTIA 400
The MTIA 400 delivers 16.9x the BF16 throughput of the MTIA 200 (3,000 vs 177 TFLOPS dense). The MTIA 400 also carries 160 GB more memory (288 GB vs 128 GB).
Meta MTIA 200
Full specs →Meta MTIA 400
Full specs →Specifications
Performance (TFLOPS)
FLOPS by Precision
What actually differs
The MTIA 400 is the newer part: MTIA, launched in 2026, against the MTIA 200's MTIA from 2024. Newer architectures typically add lower-precision formats and better throughput per watt, so check the precision rows your workload actually uses.
Memory is 128 GB against 288 GB, fed at 205 GB/s versus 9.2 TB/s. More memory per GPU means larger models fit before you have to shard across cards, which often matters more than raw TFLOPS for inference.
For LLM training and inference, weight the FP8 and FP16 rows and memory capacity most heavily. For scientific and HPC workloads, the FP64 row is the one to read.
Frequently asked questions
Is the MTIA 200 faster than the MTIA 400?
At BF16 precision the MTIA 400 reaches 3,000 TFLOPS dense against 177 TFLOPS for the MTIA 200. The performance table on this page lists every published precision for both GPUs.
Which has more memory, the MTIA 200 or the MTIA 400?
The MTIA 400 carries 288 GB of memory versus 128 GB for the MTIA 200. Memory bandwidth is 205 GB/s for the MTIA 200 and 9.2 TB/s for the MTIA 400.
How much power do the MTIA 200 and the MTIA 400 draw?
The MTIA 200 is rated at 90 W TDP and the MTIA 400 at 1.2 kW.
Get Comparison Updates
New GPUs added weekly. Be the first to see how they compare.