Laptop with RTX 4070 8GB can run Gemma 3 27B Instruct, but not the way you'd expect. Here's the change that makes it work.

Gemma 3 27B Instruct at Q4_K_M needs 25.4 GB once weights, KV cache at 8K context and framework overhead are counted. Laptop with RTX 4070 8GB offers 12 GB, leaving you 13.4 GB short. Gemma 3 12B Instruct is the same family at a smaller size and needs 11.1 GB at Q4_K_M, which your 12 GB already holds. Expect 28.1–54 tok/s (estimated).

What to do

Run Gemma 3 12B Instruct instead

Gemma 3 12B Instruct is the same family at a smaller size and needs 11.1 GB at Q4_K_M, which your 12 GB already holds.

What it needs

Pull Gemma 3 12B Instruct rather than Gemma 3 27B Instruct

What we checked

Same model family, 12B vs 27B parameters
VRAM at Q4_K_M: 11.1 GB vs 12 GB available

Other ways to get there

Move to Ryzen AI Max+ 395 mini PC, 128GB unified — 128 GB of unified memory leaves about 96 GB for a model after the OS takes its share, which holds Gemma 3 27B Instruct at Q4_K_M.
Rent a GPU by the hour instead — You want this daily, so renting is the way to try Gemma 3 27B Instruct before spending on hardware — not the long-run answer if the usage holds.
Consider waiting this market out — DDR5 kit prices are roughly 3-4x their mid-2025 level, and the shortage is expected to run into 2027. Relief is not expected before 2027-Q4. Meanwhile you have a way to run this that costs nothing extra.

What we ruled out, and why

32GB DDR5 kit (2x16GB) — More memory would let this model load, but the parts of it living in system RAM have to travel over PCIe, so it would run at roughly 1.5-2.9 tok/s — slower than reading speed. More VRAM, not more RAM, is what this needs.
64GB DDR5 kit (2x32GB) — More memory would let this model load, but the parts of it living in system RAM have to travel over PCIe, so it would run at roughly 1.5-2.9 tok/s — slower than reading speed. More VRAM, not more RAM, is what this needs.
96GB DDR5 kit (2x48GB) — More memory would let this model load, but the parts of it living in system RAM have to travel over PCIe, so it would run at roughly 1.5-2.9 tok/s — slower than reading speed. More VRAM, not more RAM, is what this needs.
128GB DDR5 kit (4x32GB) — More memory would let this model load, but the parts of it living in system RAM have to travel over PCIe, so it would run at roughly 1.5-2.9 tok/s — slower than reading speed. More VRAM, not more RAM, is what this needs.
Replace your graphics card — A laptop's graphics chip is soldered to the mainboard and cannot be replaced or added to. Renting, or a desktop, is the honest path from here.
Add a second graphics card — A laptop's graphics chip is soldered to the mainboard and cannot be replaced or added to. Renting, or a desktop, is the honest path from here.

Market conditions: DDR5 kit prices are roughly 3-4x their mid-2025 level, and the shortage is expected to run into 2027. Relief is not expected before 2027-Q4. Source, checked 2026-08-26.

Your machine is not exactly this one. Run this for your exact setup — the form opens pre-filled with Laptop with RTX 4070 8GB and Gemma 3 27B Instruct.

Questions people ask about this pairing

Can a Laptop with RTX 4070 8GB run Gemma 3 27B Instruct?

Not as it stands. Gemma 3 27B Instruct needs 25.4 GB at Q4_K_M and this machine has 12 GB available, a shortfall of 13.4 GB. Gemma 3 12B Instruct is the same family at a smaller size and needs 11.1 GB at Q4_K_M, which your 12 GB already holds.

Would more system RAM fix this?

More memory would let this model load, but the parts of it living in system RAM have to travel over PCIe, so it would run at roughly 1.5-2.9 tok/s — slower than reading speed. More VRAM, not more RAM, is what this needs.

Should I wait for prices to come down?

DDR5 kit prices are roughly 3-4x their mid-2025 level, and the shortage is expected to run into 2027. Relief is not expected before 2027-Q4. The current estimate for relief is 2027-Q4. If you can run this some other way meanwhile, waiting is defensible; if you cannot, the part still does the job today.

How fast will Gemma 3 27B Instruct actually run?

Around 28.1–54 tok/s (estimated) on this setup at Q4_K_M. Decode speed on a local model is set by memory bandwidth rather than compute, so the figure moves with the card's bandwidth and the bytes read per token — not with its price.

Does a longer context change what Gemma 3 27B Instruct needs here?

Yes, and it is the figure people forget. The 25.4 GB above already includes the KV cache at 8K; that cache grows roughly linearly with context, so doubling the window adds real gigabytes rather than a rounding error. If you plan to work with long documents on Laptop with RTX 4070 8GB, size for the context you will actually use, not the default.

Data behind this page last checked 2026-08-26.