Best local AI models for 16GB VRAM
Ranked by what actually fits at 32K context, computed from real file bytes.
A 16GB card gives you about 14.88 GiB to work with after driver overhead. 1688 indexed models fit at 32K context — the largest being Qwen3-Coder-Next-REAM at 60.3B parameters in I1-IQ1_M.
From the file· fit from summed bytesFrom the file· KV per layer
alltext generation 1435vision + language 151speech recognition 39speech synthesis 21embeddings 26video generation 15image generation 1
Fits in 16GB at 32K context
largest quantization that fits, per model
| Model | Modality | Best quant | Params○ | Total◐ | Headroom◐ |
|---|---|---|---|---|---|
| Qwen3-Coder-30B-A3B-InstructMoE | text generation | Q2_K_L | 30.5B | 14.35 GiB | 0.53 GiB |
| Qwen3.6-27B | text generation | Q3_K_S | 27.8B | 14.57 GiB | 0.31 GiB |
| Qwen3.6-35B-A3BMoE | vision + language | UD-IQ3_XXS | 36.0B | 14.53 GiB | 0.35 GiB |
| Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-MTP | vision + language | I1-IQ3_M | 27.8B | 14.75 GiB | 0.13 GiB |
| Qwen3.8-27B | text generation | Q3_K_S | 27.8B | 14.57 GiB | 0.31 GiB |
| Qwen3.5-9B | vision + language | Q8_0 | 9.7B | 10.95 GiB | 3.93 GiB |
| gemma-4-26B-A4B-itMoE | vision + language | IQ3_M | 26.5B | 14.70 GiB | 0.18 GiB |
| gemma-4-12B-it | vision + language | Q6_K_L | 12.0B | 13.08 GiB | 1.80 GiB |
| nemotron-3.5-asr-streaming-0.6b | speech recognition | F32 | 638M | 3.22 GiB | 11.66 GiB |
| Qwen3.5-4B | vision + language | BF16 | 4.7B | 9.88 GiB | 5.00 GiB |
| gemma-4-12B-it-qat-q4_0-unquantized | text generation | Q4_0 | 12.0B | 9.81 GiB | 5.07 GiB |
| Qwythos-9B-Claude-Mythos-5-1M | vision + language | Q5_K_M | 9.4B | 14.12 GiB | 0.76 GiB |
| gemma-4-E4B-it | text generation | Q8_0 | 8.0B | 8.95 GiB | 5.93 GiB |
| Qwen3-30B-A3B-Thinking-2507MoE | text generation | Q2_K_L | 30.5B | 14.35 GiB | 0.53 GiB |
| Muse-Glimmer-30B | vision + language | UD-IQ3_M | 29.8B | 14.54 GiB | 0.34 GiB |
| Qwen3-4B | text generation | BF16 | 4.0B | 12.81 GiB | 2.07 GiB |
| Qwen3-8B | text generation | Q8_0 | 8.2B | 13.44 GiB | 1.44 GiB |
| Laguna-XS-2.1MoE | text generation | Q2_K_L | 33.4B | 13.68 GiB | 1.20 GiB |
| Qwen3.5-0.8B | vision + language | BF16 | 873M | 2.60 GiB | 12.28 GiB |
| Llama-3.2-1B-Instruct | text generation | F16 | 1.2B | 4.11 GiB | 10.77 GiB |
| Qwen-AgentWorld-35B-A3BMoE | text generation | UD-IQ3_XXS | 34.7B | 14.23 GiB | 0.65 GiB |
| gemma-4-E2B-it | vision + language | BF16 | 5.1B | 9.71 GiB | 5.17 GiB |
| gpt-oss-20bMoE | text generation | F16 | 21.5B | 14.40 GiB | 0.48 GiB |
| KAT-Coder-V2.5-DevMoE | text generation | Q2_K_L | 34.7B | 13.64 GiB | 1.24 GiB |
| gemma-4-E4B-it-qat-q4_0-unquantized | vision + language | Q4_0 | 7.9B | 6.12 GiB | 8.76 GiB |
| Qwen3-VL-30B-A3B-InstructMoE | vision + language | Q2_K_L | 31.1B | 14.35 GiB | 0.53 GiB |
| Qwen3-30B-A3BMoE | text generation | Q2_K_L | 30.5B | 14.35 GiB | 0.53 GiB |
| Qwopus3.6-35B-A3B-v1MoE | vision + language | I1-IQ3_XXS | 36.0B | 14.12 GiB | 0.76 GiB |
| llama-3-youko-8b | text generation | Q8_0 | 8.0B | 12.79 GiB | 2.09 GiB |
| Qwen3.5-9B-The-Defiant-Fable-Uncensored-Heretic-NEO-IMATRIX-MAX-MTP | vision + language | Q4_K_M | 9.7B | 14.70 GiB | 0.18 GiB |
| parakeet-tdt-0.6b-v3 | speech recognition | F32 | 627M | 3.18 GiB | 11.70 GiB |
| Qwen3.5-35B-A3BMoE | vision + language | Q2_K_L | 36.0B | 14.47 GiB | 0.41 GiB |
| Llama-3.1-8B-Instruct | text generation | Q8_0 | 8.0B | 12.79 GiB | 2.09 GiB |
| ced-base | text generation | F32 | 86M | 1.16 GiB | 13.72 GiB |
| Ace-Step1.5 | speech synthesis | Q4_K | 160M | 9.65 GiB | 5.23 GiB |
| Qwen2.5-7B-Instruct | text generation | Q8_0 | 7.6B | 10.15 GiB | 4.73 GiB |
| Qwen3-TTS-12Hz-0.6B-Base | speech synthesis | Q8_0 | 915M | 8.68 GiB | 6.20 GiB |
| Qwythos-9B-v2 | vision + language | Q5_K_M | 9.7B | 14.16 GiB | 0.72 GiB |
| UI-TARS-1.5-7B | text generation | Q8_0 | 8.3B | 10.15 GiB | 4.73 GiB |
| gemma-3-1b-it | text generation | F16 | 1000M | 2.81 GiB | 12.07 GiB |
| gemma-4-E2B-it-qat-q4_0-unquantized | text generation | BF16 | 5.1B | 9.83 GiB | 5.05 GiB |
| Qwen3-1.7B | text generation | BF16 | 2.0B | 8.08 GiB | 6.80 GiB |
| GLM-4.7-FlashMoE | text generation | Q3_K_S | 31.2B | 14.84 GiB | 0.04 GiB |
| whisper-medium | speech recognition | F32 | 764M | 3.69 GiB | 11.19 GiB |
| embeddinggemma-300m | text generation | F32 | 303M | 2.05 GiB | 12.83 GiB |
| Llama-3.2-3B-Instruct | text generation | F16 | 3.2B | 10.30 GiB | 4.58 GiB |
| Qwen3-0.6B | text generation | BF16 | 752M | 5.68 GiB | 9.20 GiB |
| Qwen3-14B | text generation | Q4_1 | 14.8B | 14.60 GiB | 0.28 GiB |
| Ornith-1.0-35BMoE | text generation | UD-IQ3_XXS | 34.7B | 14.23 GiB | 0.65 GiB |
| ThinkingCap-Qwen3.6-27B | vision + language | IQ3_XXS | 27.4B | 14.62 GiB | 0.26 GiB |
| Qwopus3.6-27B-Coder | vision + language | IQ2_M | 27.8B | 12.60 GiB | 2.28 GiB |
| Voxtral-Mini-4B-Realtime-2602 | speech recognition | F16 | 4.4B | 12.33 GiB | 2.55 GiB |
| Wan2.1-T2V-1.3B | text generation | Q8_0 | 1.4B | 12.18 GiB | 2.70 GiB |
| LFM2.5-1.2B-Instruct | text generation | BF16 | 1.2B | 3.37 GiB | 11.51 GiB |
| Qwen2.5-Coder-7B-Instruct | text generation | Q6_K | 7.6B | 14.25 GiB | 0.63 GiB |
| Qwen3-VL-4B-Instruct | vision + language | BF16 | 4.4B | 12.81 GiB | 2.07 GiB |
| Qwen2.5-1.5B-Instruct | text generation | F16 | 1.5B | 4.56 GiB | 10.32 GiB |
| Jan-v3-4B-base-instruct | text generation | BF16 | 4.4B | 13.53 GiB | 1.35 GiB |
| gemma-3-4b-it | text generation | BF16 | 4.3B | 8.85 GiB | 6.03 GiB |
| whisper-large-v3 | speech recognition | F16 | 1.5B | 3.74 GiB | 11.14 GiB |
This page models a generic 16GB accelerator, so it answers what fits rather than how fast it runs. For tokens per second you need a specific card — pick one from hardware, where bandwidth is known.