Google95 GB VRAM819 GB/s$4.2/hr

Codestral 25.08 on TPU v5p (per-chip, 95GB): Does It Fit, How Fast, What It Costs

Codestral 25.08 at FP8 (E4M3) needs 22.0 GB for its weights and 37.0 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
37.0 GB needed / 95 GB available
Decode speed
36.30
tok/s per stream
Aggregate
435.6
tok/s @ 8 concurrent
Cost / 1M tokens
$2.68
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 GB37.0 GB36.30435.61.63 s$2.68✓
2× TPU190 GB37.0 GB61.701481831.5 ms$1.58✓
4× TPU380 GB37.0 GB123.45924430.7 ms$0.7878✓
8× TPU760 GB37.0 GB246.823694230.4 ms$0.3939✓
Concurrency vs context length
On 1× TPU v5p (per-chip, 95GB) at FP8 (E4M3). KV cache scales linearly with context, so this is the number of users your GPUs can actually serve.
ContextConcurrent usersKV per requestFits?
4K77962.1 MB✓ fits
8K381.88 GB✓ fits
33K97.52 GB✓ fits
66K415.03 GB✓ fits
131K230.06 GB✓ fits
Other GPUs that run Codestral 25.08
Same workload, same quantization — ranked by value (tokens per dollar-hour).
GPUGPUsAggregate tok/sCost / 1M tok
RTX 5090 32GB2×3240$0.0857
AMD MI300X 192GB1×2819$0.2464
A100 40GB SXM41×824.3$0.3033
AMD MI325X 288GB1×3191$0.3047
A100 80GB SXM41×1064$0.3134
Intel Gaudi 3 128GB1×1755$0.3166
Frequently asked questions

Can Codestral 25.08 run on TPU v5p (per-chip, 95GB)?

Yes. Codestral 25.08 at FP8 (E4M3) needs 37.0 GB including the KV cache, and TPU v5p (per-chip, 95GB) has 95 GB — so it fits on a single GPU.

How fast is Codestral 25.08 on TPU v5p (per-chip, 95GB)?

36.30 tokens/sec per stream and 435.6 tok/s aggregate with continuous batching at 8 concurrent users. Time to first token is 1.63 s, and inter-token latency is 27.6 ms.

What does Codestral 25.08 on TPU v5p (per-chip, 95GB) cost per million tokens?

$2.68 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 Codestral 25.08 need on TPU v5p (per-chip, 95GB)?

22.0 GB for the weights at FP8 (E4M3), plus 15.0 GB of KV cache for 8 concurrent requests at 8K context — 37.0 GB total. Every extra concurrent user at long context costs more, so leave headroom.

Is TPU v5p (per-chip, 95GB) the right choice for Codestral 25.08?

It works. Compare $2.68 per 1M tokens against the alternatives below — the cheapest option that fits the same workload is 4× RTX 3080 10GB at $0.0404, and the fastest is 1× Cerebras CS-3 (WSE-3) at 10636 tok/s.

Adjust the workload yourself

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

Open in calculator