Can you run Mellum 2.1 Thinking locally?

Mellum family · 2026 · 12.2B parameters (2.5B activated per token) · Hugging Face model card

How much VRAM does Mellum 2.1 Thinking need at each quantization?

Mellum 2.1 Thinking needs 8.3 GB of VRAM at Q4_K_M. The table below lists weights-only size and total VRAM including overhead for each common quantization level with 4k and 32k token contexts.

config.json declares alternating sliding_attention and full attention layers; 4 KV heads with head_dim 128 keep long-context KV cost low.

Method: weights = parameters × bits-per-weight (Q4_K_M ≈ 4.8, Q5_K_M ≈ 5.7, Q6_K ≈ 6.6, Q8_0 ≈ 8.5, FP16 = 16), plus 10% loading overhead, plus KV cache from the verified architecture config (28 layers, 4 KV heads, 128 head dim). Source: HF-verified 2026-10-12 via plain api (gated=false safetensors.total=12149923072 params=12.15B ctx=131072 license=apache-2.0 repo=JetBrains/Mellum2.1-12B-A2.5B-Thinking@92ddae9fc7665e9f801d141d2e5a6b2caf2460c4 layers=28 hidden=2304 kv_heads=4 heads=32 vocab=98304). MoE: num_experts=64 num_experts_per_tok=8 active_params_b=2.5 (vendor model card (12B total, 2.5B active, 8 activated experts); config num_experts=64 routed 8 per token). VRAM figures are documented calculations (Q4-class ~0.6 GB per B params, on active params for MoE rows), not measurements. release_date is the HF repo lastModified date observed on the verification pass. No tok/s or benchmark scores invented.

Which GPUs can run Mellum 2.1 Thinking locally?

At Q4_K_M with a 4k context, Mellum 2.1 Thinking needs 8.3 GB of VRAM. The lists below are computed live from our GPU database and grouped by how much headroom the card has. Disclosure: CompareAIHardware.com participates in the Amazon Associates program and earns from qualifying purchases through links on this page. Affiliate relationships do not influence our recommendations.

Runs Mellum 2.1 Thinking comfortably (25%+ VRAM headroom)

Minimum GPUs that fit Mellum 2.1 Thinking at Q4_K_M

These GPUs hold the model but leave little headroom — keep contexts short.

Needs 2+ GPUs to run Mellum 2.1 Thinking

One of these cards is too small on its own, but a pair (tensor or pipeline parallel, ~90% efficiency) covers the 8.3 GB requirement. See our multi-GPU guide for setup.

Can a Mac run Mellum 2.1 Thinking?

Yes — these Apple Silicon machines fit Mellum 2.1 Thinking at Q4_K_M, since macOS lets the GPU use about 75% of unified memory. Generation speed is bound by memory bandwidth, so the GB/s column matters as much as capacity.

Can Mellum 2.1 Thinking run on mini PCs, Jetson, or NPU devices?

These edge and NPU devices from our database have enough memory for Mellum 2.1 Thinking at Q4_K_M. Their memory bandwidth is far below discrete GPUs, so expect a fraction of desktop generation speed.

Which pre-built systems can run Mellum 2.1 Thinking?

These mini PCs and workstations from our database fit Mellum 2.1 Thinking at Q4_K_M. Usable-memory figures are conservative: Windows shares about half of system RAM with the GPU by default (Linux can expose more), macOS lets Apple Silicon GPUs use about 75% of unified memory, and Linux unified-memory systems such as GB10 expose roughly 90%. Generation speed is bound by memory bandwidth, so compare the GB/s column before buying.

Frequently asked questions