Google95 GB VRAM819 GB/s$4.2/hr

Mixtral 8x22B (MoE) on TPU v5p (per-chip, 95GB): Does It Fit, How Fast, What It Costs

Mixtral 8x22B (MoE) at AWQ 4-bit needs 77.6 GB for its weights and 92.6 GB in total at 8K context with 8 concurrent users. That means it fits on a single TPU v5p (per-chip, 95GB). Every figure is computed by tokcalc's open-source capacity model.

Fits?
Yes, 1 GPU
92.6 GB needed / 95 GB available
Decode speed
21.84
tok/s per stream
Aggregate
262.1
tok/s @ 8 concurrent
Cost / 1M tokens
$4.45
1× TPU v5p (per-chip, 95GB)
What changes as you add TPU v5p (per-chip, 95GB)s
Tensor-parallel degree vs VRAM, throughput, and cost. Cost scales linearly with GPU count; decode throughput scales near-linearly too, so cost per token stays roughly flat.
GPUsTotal VRAMVRAM neededDecode tok/sAggregate tok/sTTFTCost / 1M tokFits?
1× TPU95 GB92.6 GB21.84262.12.86 s$4.45✓
2× TPU190 GB92.6 GB37.13891.11.45 s$2.62✓
4× TPU380 GB92.6 GB74.263564739.5 ms$1.31✓
8× TPU760 GB92.6 GB148.514257384.8 ms$0.6546✓
Concurrency vs context length
On 1× TPU v5p (per-chip, 95GB) at AWQ 4-bit. KV cache scales linearly with context, so this is the number of users your GPUs can actually serve.
ContextConcurrent usersKV per requestFits?
4K18962.1 MB✓ fits
8K91.88 GB✓ fits
33K27.52 GB✓ fits
66K115.03 GB✓ fits
131K030.06 GB✗ does not fit
Other GPUs that run Mixtral 8x22B (MoE)
Same workload, same quantization — ranked by value (tokens per dollar-hour).
GPUGPUsAggregate tok/sCost / 1M tok
RTX 5090 32GB4×7799$0.0712
A100 80GB SXM42×2176$0.3064
H100 SXM5 80GB2×3645$0.3811
AMD MI300X 192GB1×1696$0.4095
L40S 48GB2×940.0$0.4137
AMD MI325X 288GB1×1920$0.5064
Frequently asked questions

Can Mixtral 8x22B (MoE) run on TPU v5p (per-chip, 95GB)?

Yes. Mixtral 8x22B (MoE) at AWQ 4-bit needs 92.6 GB including the KV cache, and TPU v5p (per-chip, 95GB) has 95 GB — so it fits on a single GPU.

How fast is Mixtral 8x22B (MoE) on TPU v5p (per-chip, 95GB)?

21.84 tokens/sec per stream and 262.1 tok/s aggregate with continuous batching at 8 concurrent users. Time to first token is 2.86 s, and inter-token latency is 45.8 ms.

What does Mixtral 8x22B (MoE) on TPU v5p (per-chip, 95GB) cost per million tokens?

$4.45 per 1M output tokens at 1× TPU v5p (per-chip, 95GB) ($4.20/hr). That is the marginal hardware cost — add idle capacity, ops, and egress for the real figure.

How much VRAM does Mixtral 8x22B (MoE) need on TPU v5p (per-chip, 95GB)?

77.6 GB for the weights at AWQ 4-bit, plus 15.0 GB of KV cache for 8 concurrent requests at 8K context — 92.6 GB total. Every extra concurrent user at long context costs more, so leave headroom.

Is TPU v5p (per-chip, 95GB) the right choice for Mixtral 8x22B (MoE)?

It works. Compare $4.45 per 1M tokens against the alternatives below — the cheapest option that fits the same workload is 8× RTX 4080 16GB at $0.0534, and the fastest is 4× Cerebras CS-3 (WSE-3) at 87040 tok/s.

Adjust the workload yourself

Same formulas, your context length, concurrency, and quantisation.

Open in calculator