Cheapest box to run Gemma-3-27B at BF16
Google’s Gemma-3-27B is 27B parameters, dense. The number below is computed from that and each vendor’s published specs, not asserted. Switch the quant or the context and it recomputes in place.
Gemma-3-27B at BF16 needs 57GB.
Cheapest Config that fits: Ascent GX10, GB10 Grace Blackwell, 128GB. Fits, 63GB headroom. Est. 3.3 tok/s.
Best offer today: $3,999.99 at Best Buy — in stock, checked 15 hours ago.
Collecting price history — 2 of 90 days. No timing verdict yet.
Offers — Ascent GX10, GB10 Grace Blackwell, 128GB, 1TB
- Mac mini, M6, 16GB$1,099Won't fit−44GB headroom
- Mac mini, M6, 24GB$1,299Won't fit−36GB headroom
- Mac mini, M6, 32GB$1,499Won't fit−28GB headroom
- Mac mini, M5 Pro, 24GB$1,699Won't fit−36GB headroom
- Mac mini, M5 Pro, 48GB$2,499Won't fit−13GB headroom
- Mac Studio, M5 Max, 36GB$2,499Won't fit−24GB headroom
- Mac mini, M5 Pro, 64GB$2,899Tight2GB headroomest. 3.7 tok/s
- GeForce RTX 5090, 32GB$2,999.99+hostWon't fit−27GB headroom
- Mac Studio, M5 Max, 48GB$3,099Won't fit−13GB headroom
- Mac Studio, M5 Max, 64GB$3,499Tight2GB headroomest. 7.4 tok/s
- Fits63GB headroomest. 3.3 tok/s
- Fits63GB headroomest. 3.3 tok/s
- Fits63GB headroomest. 7.4 tok/s
- Fits31GB headroomest. 14.4 tok/s
- Fits191GB headroomest. 14.4 tok/s
- Fits447GB headroomest. 14.4 tok/s
| Config | Memory | Price | $/GB | Fit | Headroom | Est. speed |
|---|---|---|---|---|---|---|
| Mac mini, M6 | 16GB | $1,099 | $69 | Won't fit | −44GB | — |
| Mac mini, M6 | 24GB | $1,299 | $54 | Won't fit | −36GB | — |
| Mac mini, M6 | 32GB | $1,499 | $47 | Won't fit | −28GB | — |
| Mac mini, M5 Pro | 24GB | $1,699 | $71 | Won't fit | −36GB | — |
| Mac mini, M5 Pro | 48GB | $2,499 | $52 | Won't fit | −13GB | — |
| Mac Studio, M5 Max | 36GB | $2,499 | $69 | Won't fit | −24GB | — |
| Mac mini, M5 Pro | 64GB | $2,899 | $45 | Tight | 2GB | 3.7 tok/s |
| GeForce RTX 5090 | 32GB | $2,999.99+host | $94 | Won't fit | −27GB | — |
| Mac Studio, M5 Max | 48GB | $3,099 | $65 | Won't fit | −13GB | — |
| Mac Studio, M5 Max | 64GB | $3,499 | $55 | Tight | 2GB | 7.4 tok/s |
| Ascent GX10, GB10 Grace Blackwell | 128GB | $3,999.99 | $31 | Fits | 63GB | 3.3 tok/s |
| DGX Spark, GB10 Grace Blackwell | 128GB | $4,699 | $37 | Fits | 63GB | 3.3 tok/s |
| Mac Studio, M5 Max | 128GB | $5,099 | $40 | Fits | 63GB | 7.4 tok/s |
| Mac Studio, M5 Ultra | 96GB | $5,499 | $57 | Fits | 31GB | 14.4 tok/s |
| Mac Studio, M5 Ultra | 256GB | $9,499 | $37 | Fits | 191GB | 14.4 tok/s |
| Mac Studio, M5 Ultra | 512GB | — | — | Fits | 447GB | 14.4 tok/s |
GeForce RTX 5090 is a card, not a computer: every price shown for it is the cheapest listing we found plus $1,000 for the minimum host machine.
memory = 27B × 2.000 B/param (16 bits) + 0.254 GB/1K × 8K K/V + 1.2 GB runtime
speed = bandwidth × 0.65 ÷ (27B active × 2.000 B/param) = 54.0 GB read per token
usable = memory − operating system reserve (8% on a Mac, 6% on a DGX Spark, 4% on a card)
Bandwidth is the vendor’s published figure. Every speed is an estimate and says so. Methodology.
What else Ascent GX10 GB10 Grace Blackwell 128GB runs
- Qwen3-235B-A22BQ3_K_M, est. 12.7 tok/s
- Mistral-Large-2411Q6_K, est. 1.8 tok/s
- gpt-oss-120bQ6_K, est. 32.6 tok/s
- Llama-4-Scout-17B-16EQ8_0, est. 7.6 tok/s
- Llama-3.3-70BQ8_0, est. 2.4 tok/s
- Qwen3-32BBF16, est. 2.8 tok/s
Best quant that still fits at 8K context, largest model first. Full list on the Config page.
Bandwidth 273GB/s, usable memory 120GB of 128GB.