model family
Hunyuan MT 2 logo

Hunyuan MT 2

Published by tencent. 2 members measured on real machines, across 2 builds. Every number below was signed by the machine that produced it — none of it is from a model card.

what it takes, what it returns
text → text

Declared by the network’s catalog, not measured — a signature says what this family would be asked. Whether it answered is the signed capability flags on each member.

2 of 54 builds measured · 52 still open
what each build proved
30b-a3b · Q4_K_M
1.8b · Q4_K_M
30b-a3b · Q4_K_M · Structured output — proved30b-a3b · Q4_K_M · Sustained output — proved30b-a3b · Q4_K_M · Writes code — proved30b-a3b · Q4_K_M · Reasons about code — proved30b-a3b · Q4_K_M · Synthesises sources — proved30b-a3b · Q4_K_M · Follows instructions — proved1.8b · Q4_K_M · Structured output — asked, failed1.8b · Q4_K_M · Sustained output — asked, failed1.8b · Q4_K_M · Writes code — asked, failed1.8b · Q4_K_M · Reasons about code — asked, failed1.8b · Q4_K_M · Synthesises sources — asked, failed1.8b · Q4_K_M · Follows instructions — asked, failed
Structured output
Sustained output
Writes code
Reasons about code
Synthesises sources
Follows instructions
Hunyuan MT 2 30B-A3B1 build measured →
hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_M30b-a3b · Q4_K_M · tencent · ollama · 2 machines · 30.1B
4Krecalled
—served
5.4best tok/s
5.4 tok/s · unnamed machine3.6 tok/s · unnamed machine
3.6 → 5.4 tok/s across 2 machines · ×1.5 spreadunnamed machine
recalled 4k⌐ declared 256k · unsignedadvertises 64× what it proved
Hunyuan MT 2 1.8B1 build measured →
hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_M1.8b · Q4_K_M · tencent · ollama · 4 machines · 1.79B
—recalled
—served
233.3best tok/s
221.8 tok/s · unnamed machine36.5 tok/s · unnamed machine14.2 tok/s · x86 · CPU · Hyderabad12.3 tok/s · x86 · CPU · Bengaluru
12.3 → 221.8 tok/s across 4 machines · ×18.0 spreadunnamed machinex86 · CPU

Builds this network knows of

Declared, not measured — this is what the catalog says exists and what it would cost a machine to hold. Nothing here is signed, and none of it claims the build works.

how much memory does your machine have

40 of 54 builds fit a 16 GB machine — budgeting 70% of it, or 11.2 GiB, because the cache and the runtime want the rest.

  • huggingface-mic/Hy-MT2-30B-A3B-micFP32vllmderived by huggingface-mic
    0.1 GiB
  • hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q2_KQ2_Kollamaquantized by mradermacher
    0.7 GiB
  • hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_SQ3_K_Sollamaquantized by mradermacher
    0.8 GiB
  • aidsoid/Hy-MT2-1.8B-MLX-4bit4BITmlxquantized by aidsoid
    0.9 GiB
  • mlx-community/Hy-MT2-1.8B-4bit4BITmlxderived by mlx-community
    0.9 GiB
  • hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_LQ3_K_Lollamaquantized by mradermacher
    0.9 GiB
  • hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_MQ4_K_Mollamameasured herepublished by tencent
    1.1 GiB
  • hf.co/tencent/Hy-MT2-1.8B-GGUF:Q6_KQ6_Kollamapublished by tencent
    1.4 GiB
  • hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:IQ1_SIQ1_Sollamaderived by mradermacher
    1.7 GiB
  • mlx-community/Hy-MT2-1.8B-8bit8BITmlxderived by mlx-community
    1.8 GiB
  • mlx-community/Hy-MT2-1.8B-Abliterated-8bit8BITmlxderived by mlx-community
    1.8 GiB
  • hf.co/tencent/Hy-MT2-1.8B-GGUF:Q8_0Q8_0ollamapublished by tencent
    1.8 GiB
  • tencent/Hy-MT2-1.8B-FP8FP8vllmpublished by tencent
    1.9 GiB
  • hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:IQ2_XXSIQ2_XXSollamaderived by mradermacher
    2.1 GiB
  • hf.co/unsloth/Hy-MT2-7B-GGUF:IQ2_MIQ2_Mollamaquantized by unsloth
    2.6 GiB
  • hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:IQ3_XXSIQ3_XXSollamaderived by mradermacher
    2.8 GiB
  • hf.co/unsloth/Hy-MT2-7B-GGUF:IQ3_XXSIQ3_XXSollamaquantized by unsloth
    2.9 GiB
  • fumetodev/Hy-MT2-1.8B-JP-Manga-Finetune-chinese-zh-Hans-v1-GGUFBF16vllmderived by fumetodev
    3.3 GiB
  • fumetodev/Hy-MT2-1.8B-JP-Manga-Finetune-v5-GGUFBF16vllmderived by fumetodev
    3.3 GiB
  • Jour/hy-mt2-1.8b-discordBF16vllmderived by Jour
    3.3 GiB
  • mlx-community/Hy-MT2-1.8BBF16mlxderived by mlx-community
    3.3 GiB
  • hf.co/mradermacher/Hy-MT2-1.8B-GGUF:F16F16ollamaquantized by mradermacher
    3.3 GiB
  • tencent/Hy-MT2-1.8BBF16vllmpublished by tencent
    3.8 GiB
  • hf.co/unsloth/Hy-MT2-7B-GGUF:IQ4_XSIQ4_XSollamaquantized by unsloth
    3.9 GiB
  • mlx-community/Hy-MT2-7B-4bit4BITmlxderived by mlx-community
    3.9 GiB
  • hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:Q4_0Q4_0ollamaderived by mradermacher
    4.1 GiB
  • hf.co/tencent/Hy-MT2-7B-GGUF:Q4_K_MQ4_K_Mollamapublished by tencent
    4.3 GiB
  • hf.co/alexwkleung/Hy-MT2-7B-Abliterated-GGUF:Q4_K_MQ4_K_Mollamaderived by alexwkleung
    4.3 GiB
  • hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:Q4_K_MQ4_K_Mollamaderived by mradermacher
    4.3 GiB
  • shawnw3i/Hy-MT2-7B-AWQAWQvllmquantized by shawnw3i
    4.4 GiB
  • Intel/Hy-MT2-7B-int4-AutoRoundINT4vllmderived by Intel
    4.4 GiB
  • hf.co/alexwkleung/Hy-MT2-7B-Abliterated-GGUF:Q5_K_MQ5_K_Mollamaderived by alexwkleung
    5.0 GiB
  • hf.co/tencent/Hy-MT2-7B-GGUF:Q6_KQ6_Kollamapublished by tencent
    5.7 GiB
  • sahilchachra/hy-mt2-7b-8bit-mlx8BITmlxquantized by sahilchachra
    7.4 GiB
  • mlx-community/Hy-MT2-7B-8bit8BITmlxderived by mlx-community
    7.4 GiB
  • mlx-community/Hy-MT2-7B-Abliterated-8bit8BITmlxderived by mlx-community
    7.4 GiB
  • hf.co/tencent/Hy-MT2-7B-GGUF:Q8_0Q8_0ollamapublished by tencent
    7.4 GiB
  • hf.co/alexwkleung/Hy-MT2-7B-Abliterated-GGUF:Q8_0Q8_0ollamaderived by alexwkleung
    7.4 GiB
  • tencent/Hy-MT2-7B-FP8FP8vllmpublished by tencent
    7.5 GiB
  • hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q2_KQ2_Kollamaquantized by mradermacher
    10.3 GiB
  • hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q3_K_SQ3_K_Sollamaquantized by mradermacher
    12.2 GiB
  • mlx-community/Hy-MT2-7B-Abliterated-bf16BF16mlxderived by mlx-community
    14.0 GiB
  • hf.co/unsloth/Hy-MT2-7B-GGUF:BF16BF16ollamaquantized by unsloth
    14.0 GiB
  • hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q3_K_LQ3_K_Lollamaquantized by mradermacher
    14.6 GiB
  • tencent/Hy-MT2-7BBF16vllmpublished by tencent
    15.0 GiB
  • AutomatosX/AX-Hy-MT2-30B-A3B-MLX-AXQ-MXFP44BITmlxderived by AutomatosX
    15.1 GiB
  • Intel/Hy-MT2-30B-A3B-int4-AutoRoundINT4vllmderived by Intel
    15.4 GiB
  • QwQbb/Hy-MT2-30B-A3B-MLX-4bit4BITmlxquantized by QwQbb
    15.8 GiB
  • hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q4_K_SQ4_K_Sollamaquantized by mradermacher
    16.0 GiB
  • hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_MQ4_K_Mollamameasured herepublished by tencent
    17.0 GiB
  • tencent/Hy-MT2-30B-A3B-FP8FP8vllmpublished by tencent
    28.5 GiB
  • QwQbb/Hy-MT2-30B-A3B-MLX-8bit8BITmlxquantized by QwQbb
    29.8 GiB
  • hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q8_0Q8_0ollamapublished by tencent
    29.8 GiB
  • hf.co/litigerking/Hy-MT2-30B-A3B-GGUF:BF16BF16ollamaquantized by litigerking
    56.0 GiB

Sizes are what the catalog declares, not what a machine measured — fitting on disk is necessary and not sufficient. Nobody here has run these on your machine, and this narrows the candidates rather than certifying one.

the same builds, by member
Hunyuan MT 2 1.8B
hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q2_KQ2_Kollama741 MiB on diskquantized by mradermacher
hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_SQ3_K_Sollama831 MiB on diskquantized by mradermacher
aidsoid/Hy-MT2-1.8B-MLX-4bit4BITmlx961 MiB on diskquantized by aidsoid
hf.co/mradermacher/Hy-MT2-1.8B-GGUF:Q3_K_LQ3_K_Lollama971 MiB on diskquantized by mradermacher
hf.co/tencent/Hy-MT2-1.8B-GGUF:Q4_K_MQ4_K_Mollama1.1 GiB on diskpublished by tencent
hf.co/tencent/Hy-MT2-1.8B-GGUF:Q6_KQ6_Kollama1.4 GiB on diskpublished by tencent
hf.co/tencent/Hy-MT2-1.8B-GGUF:Q8_0Q8_0ollama1.8 GiB on diskpublished by tencent
tencent/Hy-MT2-1.8B-FP8FP8vllm1.9 GiB on diskpublished by tencent
hf.co/mradermacher/Hy-MT2-1.8B-GGUF:F16F16ollama3.3 GiB on diskquantized by mradermacher
tencent/Hy-MT2-1.8BBF16vllm3.8 GiB on diskpublished by tencent
Hy-MT2-1.8B-JP-Manga-Finetune-chinese-zh-Hans-v1-GGUF
fumetodev/Hy-MT2-1.8B-JP-Manga-Finetune-chinese-zh-Hans-v1-GGUFBF16vllm3.3 GiB on diskderived by fumetodev
Hy-MT2-1.8B-JP-Manga-Finetune-v5-GGUF
fumetodev/Hy-MT2-1.8B-JP-Manga-Finetune-v5-GGUFBF16vllm3.3 GiB on diskderived by fumetodev
hy-mt2-1.8b-discord
Jour/hy-mt2-1.8b-discordBF16vllm3.3 GiB on diskderived by Jour
Hy-MT2-1.8B-4bit
mlx-community/Hy-MT2-1.8B-4bit4BITmlx961 MiB on diskderived by mlx-community
mlx-community/Hy-MT2-1.8B-8bit8BITmlx1.8 GiB on diskderived by mlx-community
mlx-community/Hy-MT2-1.8BBF16mlx3.3 GiB on diskderived by mlx-community
Hy-MT2-1.8B-Abliterated-8bit
mlx-community/Hy-MT2-1.8B-Abliterated-8bit8BITmlx1.8 GiB on diskderived by mlx-community
Hunyuan MT 2 30B-A3B
hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q2_KQ2_Kollama10.3 GiB on diskquantized by mradermacher
hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q3_K_SQ3_K_Sollama12.2 GiB on diskquantized by mradermacher
hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q3_K_LQ3_K_Lollama14.6 GiB on diskquantized by mradermacher
QwQbb/Hy-MT2-30B-A3B-MLX-4bit4BITmlx15.8 GiB on diskquantized by QwQbb
hf.co/mradermacher/Hy-MT2-30B-A3B-GGUF:Q4_K_SQ4_K_Sollama16.0 GiB on diskquantized by mradermacher
hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q4_K_MQ4_K_Mollama17.0 GiB on diskpublished by tencent
tencent/Hy-MT2-30B-A3B-FP8FP8vllm28.5 GiB on diskpublished by tencent
QwQbb/Hy-MT2-30B-A3B-MLX-8bit8BITmlx29.8 GiB on diskquantized by QwQbb
hf.co/tencent/Hy-MT2-30B-A3B-GGUF:Q8_0Q8_0ollama29.8 GiB on diskpublished by tencent
hf.co/litigerking/Hy-MT2-30B-A3B-GGUF:BF16BF16ollama56.0 GiB on diskquantized by litigerking
AX-Hy-MT2-30B-A3B-MLX-AXQ-MXFP4
AutomatosX/AX-Hy-MT2-30B-A3B-MLX-AXQ-MXFP44BITmlx15.1 GiB on diskderived by AutomatosX
Hy-MT2-30B-A3B-mic
huggingface-mic/Hy-MT2-30B-A3B-micFP32vllm59 MiB on diskderived by huggingface-mic
Hy-MT2-30B-A3B-int4-AutoRound
Intel/Hy-MT2-30B-A3B-int4-AutoRoundINT4vllm15.4 GiB on diskderived by Intel
Hunyuan MT 2 7B
hf.co/unsloth/Hy-MT2-7B-GGUF:IQ2_MIQ2_Mollama2.6 GiB on diskquantized by unsloth
hf.co/unsloth/Hy-MT2-7B-GGUF:IQ3_XXSIQ3_XXSollama2.9 GiB on diskquantized by unsloth
hf.co/unsloth/Hy-MT2-7B-GGUF:IQ4_XSIQ4_XSollama3.9 GiB on diskquantized by unsloth
hf.co/tencent/Hy-MT2-7B-GGUF:Q4_K_MQ4_K_Mollama4.3 GiB on diskpublished by tencent
shawnw3i/Hy-MT2-7B-AWQAWQvllm4.4 GiB on diskquantized by shawnw3i
hf.co/tencent/Hy-MT2-7B-GGUF:Q6_KQ6_Kollama5.7 GiB on diskpublished by tencent
sahilchachra/hy-mt2-7b-8bit-mlx8BITmlx7.4 GiB on diskquantized by sahilchachra
hf.co/tencent/Hy-MT2-7B-GGUF:Q8_0Q8_0ollama7.4 GiB on diskpublished by tencent
tencent/Hy-MT2-7B-FP8FP8vllm7.5 GiB on diskpublished by tencent
hf.co/unsloth/Hy-MT2-7B-GGUF:BF16BF16ollama14.0 GiB on diskquantized by unsloth
tencent/Hy-MT2-7BBF16vllm15.0 GiB on diskpublished by tencent
Hy-MT2-7B-int4-AutoRound
Intel/Hy-MT2-7B-int4-AutoRoundINT4vllm4.4 GiB on diskderived by Intel
Hy-MT2-7B-4bit
mlx-community/Hy-MT2-7B-4bit4BITmlx3.9 GiB on diskderived by mlx-community
mlx-community/Hy-MT2-7B-8bit8BITmlx7.4 GiB on diskderived by mlx-community
Hy-MT2-7B-Abliterated-bf16
hf.co/alexwkleung/Hy-MT2-7B-Abliterated-GGUF:Q4_K_MQ4_K_Mollama4.3 GiB on diskderived by alexwkleung
hf.co/alexwkleung/Hy-MT2-7B-Abliterated-GGUF:Q5_K_MQ5_K_Mollama5.0 GiB on diskderived by alexwkleung
mlx-community/Hy-MT2-7B-Abliterated-8bit8BITmlx7.4 GiB on diskderived by mlx-community
hf.co/alexwkleung/Hy-MT2-7B-Abliterated-GGUF:Q8_0Q8_0ollama7.4 GiB on diskderived by alexwkleung
mlx-community/Hy-MT2-7B-Abliterated-bf16BF16mlx14.0 GiB on diskderived by mlx-community
Hy-MT2-7B-i1-GGUF
hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:IQ1_SIQ1_Sollama1.7 GiB on diskderived by mradermacher
hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:IQ2_XXSIQ2_XXSollama2.1 GiB on diskderived by mradermacher
hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:IQ3_XXSIQ3_XXSollama2.8 GiB on diskderived by mradermacher
hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:Q4_0Q4_0ollama4.1 GiB on diskderived by mradermacher
hf.co/mradermacher/Hy-MT2-7B-i1-GGUF:Q4_K_MQ4_K_Mollama4.3 GiB on diskderived by mradermacher
ask your own gram

Your concierge can answer this from the network rather than from this page. The ask carries the operator id, so it resolves to a capability and an engine — not to a name somebody has to recognise.

What has this network measured about the Hunyuan MT 2 family (epn.family.hy-mt-2), and can my gram run any of it? See /inferences/hy-mt-2

open a gramx with this ask →epn.family.hy-mt-2