Can the AMD Instinct MI300X Accelerator run gemma-4-31B-it?
IT FITS
Yes. 9 of 9 formats evaluated fit in 192 GB, at 8,192 tokens. No published file exists for this pair, so every size is rebuilt from the pinned architecture. The best quality that fits is FP16, needing 66.1 GB and running at an estimated 66 tokens per second.
Fitting in memory is not the same as loading. Whether the runtime and version you have supports this architecture and format on this machine has not been tested here.
How long a conversation. At FP16, this pair holds 262,144 tokens of context — roughly 196,608 words — needing 107.7 GB. That is Google's own configured maximum of 262,144 tokens, not the card running out. A larger device does not extend it; a documented RoPE or YaRN extension might, and is a separate question from this one.
Every format evaluated
No published file exists for this pair, so every size is rebuilt from the pinned architecture. The Basis column says which is which for each row. Decode and first-token figures are calibrated estimates, not runs on this card.
| Format | Needs | Of 192 GB | Fits | Decode | First token | Basis |
|---|---|---|---|---|---|---|
| FP16 | 66.1 GB | 34% | yes | ~66 tok/s | ~4.0 s | reconstructed size |
| Q8_0 | 36.5 GB | 19% | yes | ~121 tok/s | ~4.0 s | reconstructed size |
| Q6_K | 28.9 GB | 15% | yes | ~153 tok/s | ~4.0 s | reconstructed size |
| Q5_K_M | 25.4 GB | 13% | yes | ~174 tok/s | ~4.0 s | reconstructed size |
| Q5_0 | 24.9 GB | 13% | yes | ~179 tok/s | ~4.0 s | reconstructed size |
| Q4_K_M | 22.2 GB | 12% | yes | ~201 tok/s | ~4.0 s | reconstructed size |
| Q4_0 | 21.1 GB | 11% | yes | ~211 tok/s | ~4.0 s | reconstructed size |
| Q3_K_M | 18.7 GB | 10% | yes | ~238 tok/s | ~4.0 s | reconstructed size |
| Q2_K | 14.9 GB | 8% | yes | ~303 tok/s | ~4.0 s | reconstructed size |
Sized at 8,192 tokens of context with the whole model resident — weights, the KV cache and the runtime reserve, offload off. Speed is only quoted for a format that fits: a rate for a configuration that cannot load is not a fact about anything.
Why the first-token figure is the same on every row
First token is modelled from the arithmetic the prompt requires, and that count does not change with the weight format — which is why it reads the same on every row. Real prefill does vary by format, because a quantized matmul is a different kernel; this model does not capture that, and the figure should be read as an order of magnitude rather than a ranking between formats.
What the labels mean
- published data
- Read from a published source — a file's byte count, a model's configuration or a manufacturer's specification — or exact arithmetic on such values. Not a measurement on a machine.
- reconstructed size
- Weight size reconstructed from the pinned architecture, because no published file exists.
- size range
- Only a lower and an upper bound are claimed for this weight size.
- estimate
- Calculated from sourced inputs by a stated method; an estimate, not a measurement.
This page fixes the context at 8,192 tokens and one device. Batch, concurrent users, KV-cache format, clusters and rental cost are all in the calculator, already set to this pairing.
Open in the calculator →Everything the AMD Instinct MI300X Accelerator runsBest models for 192 GB
gemma-4-31B-it’s architecture is read from its publisher’s own config.json at a pinned revision, and the AMD Instinct MI300X Accelerator’s 192 GB and 5300 GB/s come from the manufacturer’s specification. How far each figure can be trusted is published, per format and worst case included, on the accuracy scorecard.
Building this into your own product? A free API key returns exactly these figures, and the widget puts this answer on a product page with one script tag.
Use this answer in your own product
curl -s https://llmbottleneck.com/v1/analyze \
-H "Authorization: Bearer $LLMB_KEY" -H "content-type: application/json" \
-d '{"model":"google-gemma-4-31b-it","quantization":"FP16","context":8192,"hardware":"amd-instinct-mi300x"}'Same engine, same evidence, every field sourced. Free key in one step →
<script src="https://llmbottleneck.com/widget.js" data-model="google-gemma-4-31b-it" data-quantization="FP16" data-hardware="amd-instinct-mi300x" data-context="8192"></script>
No key needed. Unbranded, with your own buy button, on Pro and Business →