What Models?

Pick the right model for your GPU — in seconds.

Best local LLMs for 16 GB VRAM (November 2026): what fits, context and speed

With 16 GB of VRAM or unified memory, 48 models run well and 7 more are a tight fit. 52 don't fit.

Speeds assume 448 GB/s, the median memory bandwidth of the 32 GPUs and memory configurations with exactly 16 GB. Yours may be faster or slower: pick your GPU for exact numbers.

Best overall

Highest MMLU-Pro score among models that run well

  1. Phi-4 14B Q4_K_M 16K context · ~45 tok/s · MMLU 84.8
  2. Qwen3.5 9B Q8_0 230K context · ~43 tok/s · MMLU 82.5
  3. Magistral Small 24B Q4_K_M 17K context · ~29 tok/s · MMLU 82
  4. Mistral Small 3.1 24B Q4_K_M 17K context · ~29 tok/s · MMLU 81
  5. Qwen 2.5 14B Q4_K_M 41K context · ~45 tok/s · MMLU 79.9

Best for coding

Highest SWE-bench Verified score among models that run well

  1. Muse Glimmer 30B Q4_K_M 32K context · ~25 tok/s · SWE-bench 76
  2. K2-Horizon 7B Q8_0 51K context · ~42 tok/s · SWE-bench 72.4
  3. Ornith 1.5 9B Q8_0 222K context · ~42 tok/s · SWE-bench 70.6
  4. MiMo V2.6 Distill Qwen 9B Q8_0 229K context · ~42 tok/s · SWE-bench 61.1
  5. GPT-OSS 20B Q8_0 128K context · ~107 tok/s · SWE-bench 60.7

GPUs with 16 GB of memory

Unified memory (Apple Silicon, Ryzen AI Max, RTX Spark)

Upgrade options

24 GB runs 22 more models well, including Qwen3.6 27B, Qwen3.5 27B and Qwen3.5 35B A3B.

The cheapest GPUs with more than 16 GB, by launch MSRP.

As an Amazon Associate, I earn from qualifying purchases.

Every model in 16 GB

100 of 214 model configurations can run on 16 GB — sorted by quality

107 models, each counted once by the best way it runs on your hardware.

Largest that runs well Muse Glimmer 30B 29.78B params · Q4_K_M · 32K context
Fastest good-quality model Phi-3 Mini 3.8B ~132 tok/s · Q4_K_M · 37K context
? MMLU (Massive Multitask Language Understanding) measures general knowledge across 57 subjects. Higher = more capable. Sorted best-first.

Runs well 48

  • Phi-4 14B 14.7B Excellent
    MMLU 84.8
    Q4_K_M
    Context 16K Speed ~45 tok/s
    Q8_0 Tight fit · 7K context
  • Qwen3.5 9B 9B Great VisionTool useReasoning
    MMLU 82.5
    Q4_K_M
    Context 256K Speed ~67 tok/s
    Q8_0
    Context 230K Speed ~43 tok/s
  • Magistral Small 24B 23.6B Great ReasoningVisionTool use
    MMLU 82
    Q4_K_M
    Context 17K Speed ~29 tok/s
    Q8_0 Doesn't fit · needs 23.5 GB

Runs well 48

  • Phi-4 14B 14.7B Excellent
    MMLU 84.8
    Q4_K_M
    Context 16K Speed ~45 tok/s
    Q8_0 Tight fit · 7K context
  • Qwen3.5 9B 9B Great VisionTool useReasoning
    MMLU 82.5
    Q4_K_M
    Context 256K Speed ~67 tok/s
    Q8_0
    Context 230K Speed ~43 tok/s
  • Magistral Small 24B 23.6B Great ReasoningVisionTool use
    MMLU 82
    Q4_K_M
    Context 17K Speed ~29 tok/s
    Q8_0 Doesn't fit · needs 23.5 GB
  • Mistral Small 3.1 24B 23.6B Great VisionTool use
    MMLU 81
    Q4_K_M
    Context 17K Speed ~29 tok/s
    Q8_0 Doesn't fit · needs 23.5 GB
  • Qwen 2.5 14B 14.7B Great Tool use
    MMLU 79.9
    Q4_K_M
    Context 41K Speed ~45 tok/s
    Q8_0 Tight fit · 7K context
  • Qwen3.5 4B 4B Great VisionTool useReasoning
    MMLU 79.1
    Q4_K_M
    Context 256K Speed ~120 tok/s
    Q8_0
    Context 256K Speed ~82 tok/s
  • Qwen3 8B 8.2B Great ReasoningTool use
    MMLU 79
    Q4_K_M
    Context 82K Speed ~74 tok/s
    Q8_0
    Context 57K Speed ~46 tok/s
  • Gemma 4 12B 11.96B Great VisionTool useReasoning
    MMLU 77.2
    Q4_K_M
    Context 256K Speed ~55 tok/s
    Q8_0
    Context 256K Speed ~33 tok/s
  • Nemotron Nano 12B 12.6B Great ReasoningTool use
    MMLU 76
    Q4_K_M
    Context 128K Speed ~53 tok/s
    Q8_0
    Context 128K Speed ~32 tok/s
  • GPT-OSS 20B 21B Great ReasoningTool use
    MMLU 75
    Q4_K_M
    Context 128K Speed ~109 tok/s

    MoE — speed based on ~1.86 GB active experts plus per-token routing overhead (full 10.83 GB loaded in VRAM)

    Q8_0
    Context 128K Speed ~107 tok/s

    MoE — speed based on ~1.93 GB active experts plus per-token routing overhead (full 11.28 GB loaded in VRAM)

  • Qwen 2.5 7B 7.62B Good Tool use
    MMLU 74.2
    Q4_K_M
    Context 128K Speed ~79 tok/s
    Q8_0
    Context 128K Speed ~49 tok/s
  • Qwen 2.5 VL 7B 7.62B Good VisionTool use
    MMLU 74.2
    Q4_K_M
    Context 128K Speed ~79 tok/s
    Q8_0
    Context 128K Speed ~49 tok/s
  • Gemma 3 12B 11.8B Good Vision
    MMLU 74
    Q4_K_M
    Context 128K Speed ~54 tok/s
    Q8_0
    Context 71K Speed ~33 tok/s
  • Qwen 2.5 Coder 14B 14.7B Good
    MMLU 74
    Q4_K_M
    Context 41K Speed ~45 tok/s
    Q8_0 Tight fit · 7K context
  • Llama 3.2 Vision 11B 10.6B Good VisionTool use
    MMLU 73
    Q4_K_M
    Context 82K Speed ~61 tok/s
    Q8_0
    Context 45K Speed ~37 tok/s
  • DeepSeek R1 Distill Qwen 14B 14.7B Good Reasoning
    MMLU 72
    Q4_K_M
    Context 41K Speed ~45 tok/s
    Q8_0 Tight fit · 7K context
  • Devstral Small 24B 23.6B Good Tool use
    MMLU 72
    Q4_K_M
    Context 17K Speed ~29 tok/s
    Q8_0 Doesn't fit · needs 23.5 GB
  • Phi-3 Mini 3.8B 3.82B Good
    MMLU 70.9
    Q4_K_M
    Context 37K Speed ~132 tok/s
    Q8_0
    Context 33K Speed ~89 tok/s
  • Phi-4 Mini 3.8B 3.84B Good
    MMLU 70
    Q4_K_M
    Context 112K Speed ~128 tok/s
    Q8_0
    Context 100K Speed ~88 tok/s
  • Gemma 4 E4B 4.5B Good VisionTool useReasoning
    MMLU 69.4
    Q4_K_M
    Context 128K Speed ~70 tok/s
    Q8_0
    Context 128K Speed ~50 tok/s
  • Llama 3.1 8B 8.03B Good Tool use
    MMLU 69.4
    Q4_K_M
    Context 93K Speed ~76 tok/s
    Q8_0
    Context 65K Speed ~47 tok/s
    fp16 Tight fit · 8K context
  • Qwen 2.5 Coder 7B 7.62B Good
    MMLU 68
    Q4_K_M
    Context 128K Speed ~79 tok/s
    Q8_0
    Context 128K Speed ~49 tok/s
  • Mistral Nemo 12B 12.2B Good Tool use
    MMLU 68
    Q4_K_M
    Context 47K Speed ~53 tok/s
    Q8_0
    Context 20K Speed ~32 tok/s
  • Qwen 2.5 VL 3B 3.09B Fair Vision
    MMLU 65.6
    Q4_K_M
    Context 128K Speed ~153 tok/s
    Q8_0
    Context 128K Speed ~105 tok/s
  • Qwen 2.5 3B 3.09B Fair Tool use
    MMLU 65.6
    Q4_K_M
    Context 32K Speed ~153 tok/s
    Q8_0
    Context 32K Speed ~105 tok/s
  • Ministral 8B 8.02B Fair Tool use
    MMLU 65
    Q4_K_M
    Context 79K Speed ~70 tok/s
    Q8_0
    Context 58K Speed ~47 tok/s
  • SmallThinker 3B 3.09B Fair Reasoning
    MMLU 65
    Q4_K_M
    Context 32K Speed ~144 tok/s
    Q8_0
    Context 32K Speed ~99 tok/s
  • Llama 3.2 3B 3.21B Fair Tool use
    MMLU 63.4
    Q4_K_M
    Context 128K Speed ~148 tok/s
    Q8_0
    Context 119K Speed ~101 tok/s
    fp16
    Context 93K Speed ~60 tok/s
  • Gemma 3 4B 3.88B Fair Vision
    MMLU 63
    Q4_K_M
    Context 128K Speed ~128 tok/s
    Q8_0
    Context 128K Speed ~87 tok/s
  • Codestral 22B 22.2B Fair
    MMLU 62
    Q4_K_M
    Context 16K Speed ~31 tok/s
    Q8_0 Doesn't fit · needs 22.2 GB
  • Gemma 4 E2B 2.3B Fair VisionTool useReasoning
    MMLU 60
    Q4_K_M
    Context 128K Speed ~101 tok/s
    Q8_0
    Context 128K Speed ~75 tok/s
  • DeepSeek R1 Distill Qwen 7B 7.62B Fair Reasoning
    MMLU 58
    Q4_K_M
    Context 128K Speed ~79 tok/s
    Q8_0
    Context 128K Speed ~49 tok/s
  • Qwen3.5 2B 2B Fair VisionTool useReasoning
    MMLU 55.3
    Q4_K_M
    Context 256K Speed ~197 tok/s
    Q8_0
    Context 256K Speed ~149 tok/s
  • StarCoder2 15B 15.6B Basic
    MMLU 51.4
    Q4_K_M
    Context 16K Speed ~41 tok/s
    Q8_0 Tight fit · 2K context
  • DeepSeek Coder 6.7B 6.74B Basic
    MMLU 50
    Q4_K_M
    Context 16K Speed ~88 tok/s
    Q8_0
    Context 16K Speed ~55 tok/s
  • Llama 3.2 1B 1.24B Basic Tool use
    MMLU 49.3
    Q4_K_M
    Context 128K Speed ~248 tok/s
    Q8_0
    Context 128K Speed ~193 tok/s
  • Qwen3.5 0.8B 0.8B Basic VisionTool useReasoning
    MMLU 29.7
    Q4_K_M
    Context 256K Speed ~292 tok/s
    Q8_0
    Context 256K Speed ~247 tok/s
  • Nemotron 3 Nano 4B 3.97B Basic ReasoningTool use
    MMLU
    Q4_K_M
    Context 256K Speed ~113 tok/s
    Q8_0
    Context 256K Speed ~86 tok/s
  • Ornith 1.5 9B 9.65B Basic Tool useReasoning
    MMLU
    Q4_K_M
    Context 256K Speed ~66 tok/s
    Q8_0
    Context 222K Speed ~42 tok/s
  • MiMo V2.6 Distill Qwen 9B 9.41B Basic VisionTool useReasoning
    MMLU
    Q4_K_M
    Context 256K Speed ~65 tok/s
    Q8_0
    Context 229K Speed ~42 tok/s
  • LFM3 1.7B 1.73B Basic Tool useReasoning
    MMLU
    Q4_K_M
    Context 256K Speed ~214 tok/s
    Q8_0
    Context 256K Speed ~157 tok/s
  • LFM2.5 2.6B 2.7B Basic Tool use
    MMLU
    Q4_K_M
    Context 128K Speed ~167 tok/s
    Q8_0
    Context 128K Speed ~116 tok/s
  • LFM2.5 24B A2B 23.97B Basic Tool useReasoning
    MMLU
    Q4_K_M
    Context 128K Speed ~124 tok/s

    MoE — speed based on ~1.38 GB active experts plus per-token routing overhead (full 13.53 GB loaded in VRAM)

    Q8_0 Doesn't fit · needs 23.8 GB
  • Mellum2.1 12B A2.5B Thinking 12.15B Basic Tool useReasoning
    MMLU
    Q4_K_M
    Context 128K Speed ~118 tok/s

    MoE — speed based on ~1.55 GB active experts plus per-token routing overhead (full 7.52 GB loaded in VRAM)

    Q8_0
    Context 128K Speed ~94 tok/s

    MoE — speed based on ~2.48 GB active experts plus per-token routing overhead (full 12.04 GB loaded in VRAM)

  • LFM3-VL 2B 1.92B Basic VisionTool useReasoning
    MMLU
    Q4_K_M
    Context 128K Speed ~214 tok/s
    Q8_0
    Context 128K Speed ~157 tok/s
  • K2-Horizon 7B 9B Basic Tool useReasoning
    MMLU
    Q4_K_M
    Context 78K Speed ~68 tok/s
    Q8_0
    Context 51K Speed ~42 tok/s
  • Nemotron 3.5 Bolt E4B 8B Basic Tool useReasoning
    MMLU
    Q4_K_M
    Context 64K Speed ~62 tok/s
    Q8_0
    Context 64K Speed ~50 tok/s
  • Muse Glimmer 30B 29.78B Basic VisionTool useReasoning
    MMLU
    Q4_K_M
    Context 32K Speed ~25 tok/s
    Q8_0 Doesn't fit · needs 27.6 GB

Tight fit 7

Tight fit 7

  • Qwen3.6 27B 27B Excellent VisionTool useReasoning
    MMLU 86.2
    Q4_K_M
    Context 9K Speed ~26 tok/s

    Only 9K context — under the 16K needed to be practical

    Q8_0 Doesn't fit · needs 28.1 GB
    GPUs that can run this

    As an Amazon Associate, I earn from qualifying purchases.

  • Qwen3.5 27B 27B Excellent VisionTool useReasoning
    MMLU 86.1
    Q4_K_M
    Context 6K Speed ~25 tok/s

    Only 6K context — under the 16K needed to be practical

    Q8_0 Doesn't fit · needs 26.7 GB
    GPUs that can run this

    As an Amazon Associate, I earn from qualifying purchases.

  • Qwen3.8 27B 27.78B Excellent VisionTool useReasoning
    MMLU 84.3
    Q4_K_M
    Context 10K Speed ~26 tok/s

    Only 10K context — under the 16K needed to be practical

    Q8_0 Doesn't fit · needs 27.1 GB
    GPUs that can run this

    As an Amazon Associate, I earn from qualifying purchases.

  • Gemma 4 26B A4B 26B Great VisionTool useReasoning
    MMLU 82.6
    Q4_K_M
    Context 6K Speed ~95 tok/s

    Only 6K context — under the 16K needed to be practical

    MoE — speed based on ~2.44 GB active experts plus per-token routing overhead (full 15.87 GB loaded in VRAM)

    Q8_0 Doesn't fit · needs 25.0 GB
    GPUs that can run this

    As an Amazon Associate, I earn from qualifying purchases.

  • Gemma 3 27B 27B Great Vision
    MMLU 78
    Q4_K_M
    Context 7K Speed ~26 tok/s

    Only 7K context — under the 16K needed to be practical

    Q8_0 Doesn't fit · needs 26.8 GB
    GPUs that can run this

    As an Amazon Associate, I earn from qualifying purchases.

  • Gemma 2 27B 27.2B Great
    MMLU 75.2
    Q4_K_M
    Context 1K Speed ~25 tok/s

    Only 1K context — under the 16K needed to be practical

    Q8_0 Doesn't fit · needs 27.3 GB
  • Gemma 2 9B 9.24B Good
    MMLU 71.3
    Q4_K_M
    Context 8K Speed ~66 tok/s

    Only 8K context — under the 16K needed to be practical

    Q8_0
    Context 8K Speed ~41 tok/s

    Only 8K context — under the 16K needed to be practical

Notify me when new models fit my GPU

An email when a new model fits your 16 GB setup — no spam, unsubscribe anytime.

Doesn't fit 52

  • DeepSeek R1 0528 671B Excellent Reasoning
    MMLU 90.8
    Q4_K_M Needs 381.2 GB — 365.2 GB over
    Q8_0 Needs 664.4 GB — 648.4 GB over
  • Qwen3 Next 80B-A3B 80B Excellent Tool use
    MMLU 90
    Q4_K_M Needs 45.1 GB — 29.1 GB over
    Q8_0 Needs 79.0 GB — 63.0 GB over
    GPUs that can run this

    As an Amazon Associate, I earn from qualifying purchases.

  • MiMo V2.5 Pro 1020B Excellent Reasoning
    MMLU 89.4
    Q4_K_M Needs 586.5 GB — 570.5 GB over
    Q8_0 Needs 1013.0 GB — 997.0 GB over

Doesn't fit 52