Local AI Advisor

New open models

Fresh releases from 25 labs that publish open weights, rated by our engine for general chat on 16 GB laptop, 24 GB GPU, 48 GB Mac and 128 GB AI PC.

RSS feed

40 of 40

  • DeepSeek-V4.1-Flash-1017

    deepseek-ai · Oct 17, 2026

    763BVisionOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Does Not Fit
    48 GB Mac
    Does Not Fit
    128 GB AI PC
    Does Not Fit
    Check on my computer Hugging Face 1,314

  • NVIDIA-Nemotron-Labs-Dory1.0-4B-MXFP8

    nvidia · Oct 16, 2026

    4B
    16 GB laptop
    Comfortable
    20–35 tok/s
    24 GB GPU
    Excellent
    150–250 tok/s
    48 GB Mac
    Excellent
    40–70 tok/s
    128 GB AI PC
    Excellent
    50–80 tok/s
    Check on my computer Hugging Face 4

  • Muse-Spark-Open-816B-A42B-DFlash2

    nvidia · Oct 16, 2026 · fine-tune of meta-models/Muse-Spark-Open-816B-A42B

    4.4BVision
    16 GB laptop
    Comfortable
    20–35 tok/s
    24 GB GPU
    Excellent
    140–240 tok/s
    48 GB Mac
    Excellent
    40–70 tok/s
    128 GB AI PC
    Excellent
    45–75 tok/s
    Check on my computer Hugging Face 5

  • Hy-Memory-30B-A3B

    tencent · Oct 15, 2026

    30BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    310–520 tok/s
    48 GB Mac
    Excellent
    75–120 tok/s
    128 GB AI PC
    Excellent
    85–140 tok/s
    Check on my computer Hugging Face 187

  • granite-switch-4.1-8b

    ibm-granite · Oct 14, 2026

    8.8BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    65–110 tok/s
    48 GB Mac
    Excellent
    18–30 tok/s
    128 GB AI PC
    Excellent
    20–35 tok/s
    Check on my computer Hugging Face 26

  • granite-switch-4.1-30b

    ibm-granite · Oct 14, 2026

    30BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    35–55 tok/s
    48 GB Mac
    Acceptable
    7–11 tok/s
    128 GB AI PC
    Acceptable
    7–12 tok/s
    Check on my computer Hugging Face 21

  • UserRepGen-4B

    microsoft · Oct 14, 2026 · fine-tune of Qwen/Qwen3.5-4B

    4.5BOpen source
    16 GB laptop
    Comfortable
    25–40 tok/s
    24 GB GPU
    Excellent
    140–230 tok/s
    48 GB Mac
    Excellent
    40–65 tok/s
    128 GB AI PC
    Excellent
    45–75 tok/s
    Check on my computer Hugging Face 10

  • Ling-3.1-flash

    inclusionAI · Oct 13, 2026

    572BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Does Not Fit
    48 GB Mac
    Does Not Fit
    128 GB AI PC
    Does Not Fit
    Check on my computer Hugging Face 121

  • LFM2.5-24B-A2B

    LiquidAI · Oct 12, 2026 · fine-tune of LiquidAI/LFM2.5-24B-A2B-Base

    24B
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    440–740 tok/s
    48 GB Mac
    Excellent
    110–190 tok/s
    128 GB AI PC
    Excellent
    120–200 tok/s
    Check on my computer Hugging Face 469

  • LFM2.5-24B-A2B-Base

    LiquidAI · Oct 9, 2026

    24B
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    400–670 tok/s
    48 GB Mac
    Excellent
    110–190 tok/s
    128 GB AI PC
    Excellent
    130–210 tok/s
    Check on my computer Hugging Face 36

  • Youtu-Parsing-Omni

    tencent · Oct 9, 2026

    5.3BVision
    16 GB laptop
    Comfortable
    25–40 tok/s
    24 GB GPU
    Excellent
    110–190 tok/s
    48 GB Mac
    Excellent
    30–55 tok/s
    128 GB AI PC
    Excellent
    35–60 tok/s
    Check on my computer Hugging Face 112

  • MiniCPM-V-4.7-35B-A3B

    openbmb · Oct 7, 2026

    35BVisionOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    180–300 tok/s
    48 GB Mac
    Excellent
    45–75 tok/s
    128 GB AI PC
    Excellent
    45–75 tok/s
    Check on my computer Hugging Face 1,005

  • MiniCPM-V-4.7-1B

    openbmb · Oct 6, 2026

    1.6BVisionOpen source
    16 GB laptop
    Comfortable
    50–80 tok/s
    24 GB GPU
    Excellent
    590–990 tok/s
    48 GB Mac
    Excellent
    110–180 tok/s
    128 GB AI PC
    Excellent
    190–310 tok/s
    Check on my computer Hugging Face 385

  • d1-omni-600M

    LiquidAI · Oct 5, 2026 · fine-tune of LiquidAI/LFM2.5-Encoder-350M

    0.6BVision
    16 GB laptop
    Excellent
    90–150 tok/s
    24 GB GPU
    Excellent
    1400–2330 tok/s
    48 GB Mac
    Excellent
    270–440 tok/s
    128 GB AI PC
    Excellent
    360–600 tok/s
    Check on my computer Hugging Face 137

  • d1-3B

    LiquidAI · Oct 5, 2026 · fine-tune of LiquidAI/LFM2.5-VL-3B

    3.1BVision
    16 GB laptop
    Comfortable
    25–40 tok/s
    24 GB GPU
    Excellent
    230–390 tok/s
    48 GB Mac
    Excellent
    55–95 tok/s
    128 GB AI PC
    Excellent
    75–120 tok/s
    Check on my computer Hugging Face 448

  • AesCode-32B

    microsoft · Sep 29, 2026 · fine-tune of Qwen/Qwen3-VL-32B-Instruct

    33BVisionOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Excellent
    30–50 tok/s
    48 GB Mac
    Acceptable
    7–12 tok/s
    128 GB AI PC
    Acceptable
    7–11 tok/s
    Check on my computer Hugging Face 53

  • AesCode-8B

    microsoft · Sep 29, 2026 · fine-tune of Qwen/Qwen3-VL-8B-Instruct

    8.8BVisionOpen source
    16 GB laptop
    Borderline
    14–25 tok/s
    24 GB GPU
    Excellent
    65–110 tok/s
    48 GB Mac
    Excellent
    18–30 tok/s
    128 GB AI PC
    Excellent
    20–35 tok/s
    Check on my computer Hugging Face 73

  • MiMo-V2.6-Flash-MOPD

    XiaomiMiMo · Sep 27, 2026

    311BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Does Not Fit
    48 GB Mac
    Does Not Fit
    128 GB AI PC
    Does Not Fit
    Check on my computer Hugging Face 65

  • MiMo-V2.6-Pro-MOPD

    XiaomiMiMo · Sep 27, 2026

    1024BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Does Not Fit
    48 GB Mac
    Does Not Fit
    128 GB AI PC
    Does Not Fit
    Check on my computer Hugging Face 38

  • Apertus-v1.5-70B-base

    swiss-ai · Sep 23, 2026

    72BVisionOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Borderline
    2–3.3 tok/s
    48 GB Mac
    Borderline
    4.5–7 tok/s
    128 GB AI PC
    Acceptable
    4.6–8 tok/s
    Check on my computer Hugging Face 3

  • Apertus-v1.5-8B-base

    swiss-ai · Sep 23, 2026

    8.8BVisionOpen source
    16 GB laptop
    Borderline
    14–25 tok/s
    24 GB GPU
    Excellent
    65–110 tok/s
    48 GB Mac
    Excellent
    18–30 tok/s
    128 GB AI PC
    Excellent
    20–35 tok/s
    Check on my computer Hugging Face 3

  • MiMo-V2.6-Distill-Qwen-9B

    XiaomiMiMo · Sep 21, 2026 · fine-tune of Qwen/Qwen3.5-9B

    9.4BVisionOpen source
    16 GB laptop
    Borderline
    12–20 tok/s
    24 GB GPU
    Excellent
    65–110 tok/s
    48 GB Mac
    Excellent
    19–30 tok/s
    128 GB AI PC
    Excellent
    20–35 tok/s
    Check on my computer Hugging Face 659

  • MiMo-V2.6-Flash-RL

    XiaomiMiMo · Sep 21, 2026

    311BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Does Not Fit
    48 GB Mac
    Does Not Fit
    128 GB AI PC
    Does Not Fit
    Check on my computer Hugging Face 565

  • MiMo-V2.6-Pro-RL

    XiaomiMiMo · Sep 21, 2026

    1024BOpen source
    16 GB laptop
    Does Not Fit
    24 GB GPU
    Does Not Fit
    48 GB Mac
    Does Not Fit
    128 GB AI PC
    Does Not Fit
    Check on my computer Hugging Face 747

  • Simple-Attention-Sparsification

    tencent · Sep 14, 2026 · fine-tune of Qwen/Qwen3-14B

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 14

  • AstaBrief_8B_SFT

    allenai · Sep 10, 2026 · fine-tune of Qwen/Qwen3-8B

    Open source

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 2

  • DeepSeek-V4.1-Flash

    deepseek-ai · Sep 10, 2026

    763BVisionOpen source

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 4,634

  • tiny-aya-base-32K

    CohereLabs · Sep 8, 2026

    3.4BNon-commercial Gated

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 4

  • JustRL-II-base-model

    openbmb · Sep 7, 2026

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 8

  • tiny-aya-en-thinker

    CohereLabs · Sep 2, 2026

    3.4BNon-commercial Gated

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 12

  • tiny-aya-l2-thinker

    CohereLabs · Sep 2, 2026

    3.4BNon-commercial Gated

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 15

  • DeepSeek-V4-Flash-Vision-Exp

    deepseek-ai · Aug 31, 2026

    305BVisionOpen source

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 955

  • Hy4-preview

    tencent · Aug 27, 2026

    780BOpen source

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 508

  • Qwen-Drive-1.0-4B

    Qwen · Aug 27, 2026 · fine-tune of Qwen/Qwen3.5-4B

    4.5BVisionOpen source

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 250

  • ContextPilot-E4B

    tencent · Aug 27, 2026 · fine-tune of google/gemma-4-E4B-it

    7.9B

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 9

  • ContextPilot-14B

    tencent · Aug 27, 2026 · fine-tune of Qwen/Qwen3-14B

    15B

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 23

  • ContextPilot-8B

    tencent · Aug 27, 2026 · fine-tune of Qwen/Qwen3-8B

    8.2B

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 11

  • Llama-3-Blama-8B

    allenai · Aug 26, 2026 · fine-tune of allenai/Llama-3-Blama-8B-Stage1

    8.3BOpen weights (restricted)

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 1

  • Llama-3-Blama-8B-Stage1

    allenai · Aug 26, 2026 · fine-tune of meta-llama/Meta-Llama-3-8B

    8.3BOpen weights (restricted)

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 0

  • Bwen-8B

    allenai · Aug 26, 2026 · fine-tune of allenai/Bwen-8B-Stage1

    8.3BOpen source

    Rating pending. Open it to rate it live.

    Check on my computer Hugging Face 2

We list each lab’s own text and vision-language repositories from the last four months and skip re-packaged copies (GGUF, AWQ, FP8 and similar). Ratings assume a typical setup (Open WebUI, normal background apps); open any model to rate it for your exact machine. Licenses are as declared on the model card.