model family
gpt-oss logo

gpt-oss

Published by openai. 6 members measured on real machines, across 13 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.

13 of 64 builds measured · 51 still open
what each build proved
20b · Q4_K_M
20b · Q8_0 · unsloth
20b · Q3_K_S
20b · Q4_0
20b · Q2_K
20b · Q8_0 · unsloth
20b~bartowski-openai-experimental · MXFP4_MOE
20b~huihui-ai-abliterated · Q8_0
20b~p-e-w-heretic · Q8_0
20b~arliai-derestricted · Q5_K_M
20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M
20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S
20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0
20b · Q4_K_M · Tool calling — proved20b · Q4_K_M · Tool loop finished — proved20b · Q4_K_M · Faithful to tool results — proved20b · Q4_K_M · Chained tool calls — proved20b · Q4_K_M · Structured output — proved20b · Q4_K_M · Sustained output — proved20b · Q4_K_M · Extended thinking — proved20b · Q4_K_M · Writes code — proved20b · Q4_K_M · Reasons about code — proved20b · Q4_K_M · Synthesises sources — proved20b · Q4_K_M · Follows instructions — proved20b · Q8_0 · unsloth · Tool calling — proved20b · Q8_0 · unsloth · Tool loop finished — proved20b · Q8_0 · unsloth · Faithful to tool results — proved20b · Q8_0 · unsloth · Chained tool calls — proved20b · Q8_0 · unsloth · Structured output — proved20b · Q8_0 · unsloth · Sustained output — proved20b · Q8_0 · unsloth · Extended thinking — proved20b · Q8_0 · unsloth · Writes code — proved20b · Q8_0 · unsloth · Reasons about code — proved20b · Q8_0 · unsloth · Synthesises sources — proved20b · Q8_0 · unsloth · Follows instructions — proved20b · Q3_K_S · Tool calling — proved20b · Q3_K_S · Tool loop finished — asked, failed20b · Q3_K_S · Faithful to tool results — proved20b · Q3_K_S · Chained tool calls — proved20b · Q3_K_S · Structured output — proved20b · Q3_K_S · Sustained output — proved20b · Q3_K_S · Extended thinking — proved20b · Q3_K_S · Writes code — proved20b · Q3_K_S · Reasons about code — proved20b · Q3_K_S · Synthesises sources — proved20b · Q3_K_S · Follows instructions — proved20b · Q4_0 · Tool calling — proved20b · Q4_0 · Tool loop finished — proved20b · Q4_0 · Faithful to tool results — proved20b · Q4_0 · Chained tool calls — never asked20b · Q4_0 · Structured output — proved20b · Q4_0 · Sustained output — proved20b · Q4_0 · Extended thinking — proved20b · Q4_0 · Writes code — proved20b · Q4_0 · Reasons about code — proved20b · Q4_0 · Synthesises sources — proved20b · Q4_0 · Follows instructions — proved20b · Q2_K · Tool calling — proved20b · Q2_K · Tool loop finished — proved20b · Q2_K · Faithful to tool results — proved20b · Q2_K · Chained tool calls — proved20b · Q2_K · Structured output — proved20b · Q2_K · Sustained output — proved20b · Q2_K · Extended thinking — proved20b · Q2_K · Writes code — proved20b · Q2_K · Reasons about code — proved20b · Q2_K · Synthesises sources — proved20b · Q2_K · Follows instructions — proved20b · Q8_0 · unsloth · Tool calling — proved20b · Q8_0 · unsloth · Tool loop finished — proved20b · Q8_0 · unsloth · Faithful to tool results — proved20b · Q8_0 · unsloth · Chained tool calls — asked, failed20b · Q8_0 · unsloth · Structured output — proved20b · Q8_0 · unsloth · Sustained output — proved20b · Q8_0 · unsloth · Extended thinking — proved20b · Q8_0 · unsloth · Writes code — proved20b · Q8_0 · unsloth · Reasons about code — proved20b · Q8_0 · unsloth · Synthesises sources — proved20b · Q8_0 · unsloth · Follows instructions — proved20b~bartowski-openai-experimental · MXFP4_MOE · Tool calling — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Tool loop finished — never asked20b~bartowski-openai-experimental · MXFP4_MOE · Faithful to tool results — never asked20b~bartowski-openai-experimental · MXFP4_MOE · Chained tool calls — never asked20b~bartowski-openai-experimental · MXFP4_MOE · Structured output — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Sustained output — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Extended thinking — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Writes code — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Reasons about code — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Synthesises sources — asked, failed20b~bartowski-openai-experimental · MXFP4_MOE · Follows instructions — asked, failed20b~huihui-ai-abliterated · Q8_0 · Tool calling — asked, failed20b~huihui-ai-abliterated · Q8_0 · Tool loop finished — never asked20b~huihui-ai-abliterated · Q8_0 · Faithful to tool results — never asked20b~huihui-ai-abliterated · Q8_0 · Chained tool calls — never asked20b~huihui-ai-abliterated · Q8_0 · Structured output — asked, failed20b~huihui-ai-abliterated · Q8_0 · Sustained output — asked, failed20b~huihui-ai-abliterated · Q8_0 · Extended thinking — asked, failed20b~huihui-ai-abliterated · Q8_0 · Writes code — asked, failed20b~huihui-ai-abliterated · Q8_0 · Reasons about code — asked, failed20b~huihui-ai-abliterated · Q8_0 · Synthesises sources — asked, failed20b~huihui-ai-abliterated · Q8_0 · Follows instructions — asked, failed20b~p-e-w-heretic · Q8_0 · Tool calling — asked, failed20b~p-e-w-heretic · Q8_0 · Tool loop finished — never asked20b~p-e-w-heretic · Q8_0 · Faithful to tool results — never asked20b~p-e-w-heretic · Q8_0 · Chained tool calls — never asked20b~p-e-w-heretic · Q8_0 · Structured output — asked, failed20b~p-e-w-heretic · Q8_0 · Sustained output — asked, failed20b~p-e-w-heretic · Q8_0 · Extended thinking — asked, failed20b~p-e-w-heretic · Q8_0 · Writes code — asked, failed20b~p-e-w-heretic · Q8_0 · Reasons about code — asked, failed20b~p-e-w-heretic · Q8_0 · Synthesises sources — asked, failed20b~p-e-w-heretic · Q8_0 · Follows instructions — asked, failed20b~arliai-derestricted · Q5_K_M · Tool calling — asked, failed20b~arliai-derestricted · Q5_K_M · Tool loop finished — never asked20b~arliai-derestricted · Q5_K_M · Faithful to tool results — never asked20b~arliai-derestricted · Q5_K_M · Chained tool calls — never asked20b~arliai-derestricted · Q5_K_M · Structured output — asked, failed20b~arliai-derestricted · Q5_K_M · Sustained output — asked, failed20b~arliai-derestricted · Q5_K_M · Extended thinking — asked, failed20b~arliai-derestricted · Q5_K_M · Writes code — asked, failed20b~arliai-derestricted · Q5_K_M · Reasons about code — asked, failed20b~arliai-derestricted · Q5_K_M · Synthesises sources — asked, failed20b~arliai-derestricted · Q5_K_M · Follows instructions — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Tool calling — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Tool loop finished — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Faithful to tool results — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Chained tool calls — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Structured output — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Sustained output — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Extended thinking — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Writes code — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Reasons about code — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Synthesises sources — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · Follows instructions — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Tool calling — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Tool loop finished — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Faithful to tool results — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Chained tool calls — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Structured output — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Sustained output — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Extended thinking — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Writes code — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Reasons about code — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Synthesises sources — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · Follows instructions — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Tool calling — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Tool loop finished — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Faithful to tool results — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Chained tool calls — never asked20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Structured output — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Sustained output — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Extended thinking — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Writes code — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Reasons about code — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Synthesises sources — asked, failed20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · Follows instructions — asked, failed
Tool calling
Tool loop finished
Faithful to tool results
Chained tool calls
Structured output
Sustained output
Extended thinking
Writes code
Reasons about code
Synthesises sources
Follows instructions
gpt-oss 20B6 builds measured →
hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_M20b · Q4_K_M · unsloth · ollama · 6 machines · 20.9B
128Krecalled
—served
73.2best tok/s
73.2 tok/s · Apple silicon · Bhubaneswar52.4 tok/s · unnamed machine24.0 tok/s · unnamed machine4.4 tok/s · unnamed machine4.3 tok/s · unnamed machine4.1 tok/s · x86 · CPU · Sydney
4.1 → 73.2 tok/s across 6 machines · ×17.9 spreadApple siliconunnamed machinex86 · CPU
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XL20b · Q8_0 · unsloth · ollama · 2 machines · 20.9B
128Krecalled
—served
44.2best tok/s
44.2 tok/s · unnamed machine3.1 tok/s · x86 · CPU · Sydney
3.1 → 44.2 tok/s across 2 machines · ×14.2 spreadunnamed machinex86 · CPU
recalled 128k⌐ declared 128k · unsigned
hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_S20b · Q3_K_S · unsloth · ollama · 3 machines · 20.9B
64Krecalled
—served
54.7best tok/s
54.7 tok/s · unnamed machine4.4 tok/s · x86 · CPU · Sydney4.3 tok/s · unnamed machine
4.3 → 54.7 tok/s across 3 machines · ×12.7 spreadunnamed machinex86 · CPU
recalled 64k⌐ declared 128k · unsignedadvertises 2× what it proved
hf.co/unsloth/gpt-oss-20b-GGUF:Q4_020b · Q4_0 · unsloth · ollama · 3 machines · 20.9B
4Krecalled
—served
4.6best tok/s
recalled 4k⌐ declared 128k · unsignedadvertises 32× what it proved
hf.co/unsloth/gpt-oss-20b-GGUF:Q2_K20b · Q2_K · unsloth · ollama · 2 machines · 20.9B
4Krecalled
—served
4.3best tok/s
recalled 4k⌐ declared 128k · unsignedadvertises 32× what it proved
hf.co/unsloth/gpt-oss-20b-GGUF:Q8_020b · Q8_0 · unsloth · ollama · 2 machines · 20.9B
4Krecalled
—served
4.0best tok/s
recalled 4k⌐ declared 128k · unsignedadvertises 32× what it proved
openai_gpt-oss-20b-GGUF-MXFP4-Experimental1 build measured →
hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP420b~bartowski-openai-experimental · MXFP4_MOE · bartowski · ollama · 1 machine · 20.9B
—recalled
—served
4.3best tok/s
Huihui-gpt-oss-20b-BF16-abliterated1 build measured →
hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_020b~huihui-ai-abliterated · Q8_0 · DavidAU · ollama · 1 machine · 20.9B
—recalled
—served
3.3best tok/s
gpt-oss-20b-heretic1 build measured →
hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_020b~p-e-w-heretic · Q8_0 · DavidAU · ollama · 1 machine · 20.9B
—recalled
—served
3.9best tok/s
gpt-oss-20b-Derestricted1 build measured →
hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_M20b~arliai-derestricted · Q5_K_M · MaziyarPanahi · ollama · 1 machine · 20.9B
—recalled
—served
3.4best tok/s
OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored3 builds measured →
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_M20b~davidau-openai-claude-4-5-opus-heretic-uncensored · IQ1_M · mradermacher · ollama · 1 machine · 20.9B
—recalled
—served
4.2best tok/s
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_S20b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q3_K_S · mradermacher · ollama · 1 machine · 20.9B
—recalled
—served
4.7best tok/s
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_020b~davidau-openai-claude-4-5-opus-heretic-uncensored · Q4_0 · mradermacher · ollama · 1 machine · 20.9B
—recalled
—served
4.8best tok/s

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

18 of 64 builds fit a 16 GB machine — budgeting 70% of it, or 11.2 GiB, because the cache and the runtime want the rest.

  • yujiepan/gpt-oss-tiny-randomBF16vllmderived by yujiepan
    0.0 GiB
  • z-lab/gpt-oss-120b-DFlashBF16vllmderived by z-lab
    1.5 GiB
  • zhuyksir/EAGLE3-gpt-oss-20b-bf16BF16vllmderived by zhuyksir
    1.6 GiB
  • nvidia/gpt-oss-120b-Eagle3-short-contextBF16vllmderived by nvidia
    1.9 GiB
  • manjunathshiva/gpt-oss-20b-tq33BITmlxderived by manjunathshiva
    9.3 GiB
  • lmstudio-community/gpt-oss-safeguard-20b-MLX-MXFP44BITmlxquantized by lmstudio-community
    10.4 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_SQ3_K_Sollamameasured herequantized by unsloth
    10.7 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:Q2_KQ2_Kollamameasured herequantized by unsloth
    10.7 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0Q4_0ollamameasured herequantized by unsloth
    10.7 GiB
  • hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:IQ2_XSIQ2_XSollamaderived by bartowski
    10.7 GiB
  • hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:Q4_0Q4_0ollamaderived by bartowski
    10.7 GiB
  • hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:Q3_K_SQ3_K_Sollamaderived by bartowski
    10.8 GiB
  • mlx-community/gpt-oss-20b-OptiQ-4bit4BITmlxderived by mlx-community
    10.8 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_MQ4_K_Mollamameasured herequantized by unsloth
    10.8 GiB
  • hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:Q4_K_MQ4_K_Mollamaderived by bartowski
    10.9 GiB
  • hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_SIQ1_Sollamaderived by mradermacher
    11.2 GiB
  • hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ2_XXSIQ2_XXSollamaderived by mradermacher
    11.2 GiB
  • hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_MIQ1_Mollamameasured herederived by mradermacher
    11.2 GiB
  • hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_SQ3_K_Sollamameasured herederived by mradermacher
    11.2 GiB
  • hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q2_KQ2_Kollamaderived by MaziyarPanahi
    11.2 GiB
  • mlx-community/gpt-oss-20b-MXFP4-Q84BITmlxquantized by mlx-community
    11.2 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0Q8_0ollamameasured herequantized by unsloth
    11.3 GiB
  • hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4MXFP4ollamameasured herederived by bartowski
    11.3 GiB
  • hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0Q4_0ollamameasured herederived by mradermacher
    11.3 GiB
  • nightmedia/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4-q8-hi-mlx4BITmlxderived by nightmedia
    11.4 GiB
  • unsloth/gpt-oss-20b-unsloth-bnb-4bitBITSANDBYTESvllmderived by unsloth
    11.7 GiB
  • hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:IQ4_NLIQ4_NLollamaderived by DavidAU
    11.8 GiB
  • hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:IQ4_NLIQ4_NLollamaderived by DavidAU
    11.8 GiB
  • hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q3_K_MQ3_K_Mollamaderived by MaziyarPanahi
    12.0 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XLQ8_K_XLollamameasured herequantized by unsloth
    12.3 GiB
  • hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q3_K_LQ3_K_Lollamaderived by MaziyarPanahi
    12.4 GiB
  • huihui-ai/Huihui-gpt-oss-20b-mxfp4-abliterated-v2MXFP4vllmderived by huihui-ai
    12.8 GiB
  • openai/gpt-oss-safeguard-20bMXFP4vllmpublished by openai
    12.8 GiB
  • hf.co/unsloth/gpt-oss-20b-GGUF:F16F16ollamaquantized by unsloth
    12.8 GiB
  • hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q4_K_MQ4_K_Mollamaderived by MaziyarPanahi
    14.7 GiB
  • hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_K_MQ4_K_Mollamaderived by mradermacher
    14.7 GiB
  • hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_MQ5_K_Mollamameasured herederived by MaziyarPanahi
    15.7 GiB
  • halley-ai/gpt-oss-20b-MLX-6bit-gs326BITmlxderived by halley-ai
    17.1 GiB
  • hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0Q8_0ollamameasured herederived by DavidAU
    20.6 GiB
  • hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0Q8_0ollamameasured herederived by DavidAU
    20.6 GiB
  • lmstudio-community/gpt-oss-20b-MLX-8bit8BITmlxquantized by lmstudio-community
    20.7 GiB
  • openai/gpt-oss-20bMXFP4vllmpublished by openai
    25.6 GiB
  • unsloth/gpt-oss-20b-BF16BF16vllmderived by unsloth
    39.0 GiB
  • hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:BF16BF16ollamaderived by bartowski
    39.0 GiB
  • OpenGVLab/InternVL3_5-GPT-OSS-20B-A4B-Preview-HFBF16vllmderived by OpenGVLab
    39.5 GiB
  • lmstudio-community/gpt-oss-safeguard-120b-MLX-MXFP44BITmlxquantized by lmstudio-community
    57.9 GiB
  • unsloth/gpt-oss-120b-GGUF:Q3_K_SQ3_K_Sllamacppquantized by unsloth
    58.3 GiB
  • unsloth/gpt-oss-120b-GGUF:Q2_KQ2_Kllamacppquantized by unsloth
    58.3 GiB
  • unsloth/gpt-oss-120b-GGUF:Q4_0Q4_0llamacppquantized by unsloth
    58.3 GiB
  • unsloth/gpt-oss-120b-GGUF:Q4_K_MQ4_K_Mllamacppquantized by unsloth
    58.5 GiB
  • mlx-community/gpt-oss-120b-MXFP4-Q84BITmlxquantized by mlx-community
    59.0 GiB
  • unsloth/gpt-oss-120b-GGUF:Q8_0Q8_0llamacppquantized by unsloth
    59.0 GiB
  • hf.co/ggml-org/gpt-oss-120b-GGUF:MXFP4MXFP4ollamaquantized by ggml-org
    59.0 GiB
  • cloudyu/gpt-oss-120b-Fable-5-Distilled4BITmlxderived by cloudyu
    59.0 GiB
  • hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:MXFP4MXFP4ollamaderived by mradermacher
    59.0 GiB
  • unsloth/gpt-oss-120b-GGUF:Q8_K_XLQ8_K_XLllamacppquantized by unsloth
    60.0 GiB
  • openai/gpt-oss-safeguard-120bMXFP4vllmpublished by openai
    60.8 GiB
  • hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q3_K_SQ3_K_Sollamaderived by mradermacher
    61.6 GiB
  • hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q2_KQ2_Kollamaderived by mradermacher
    61.6 GiB
  • hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q4_K_SQ4_K_Sollamaderived by mradermacher
    75.4 GiB
  • hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q4_K_MQ4_K_Mollamaderived by mradermacher
    81.8 GiB
  • lmstudio-community/gpt-oss-120b-MLX-8bit8BITmlxquantized by lmstudio-community
    115.6 GiB
  • hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q8_0Q8_0ollamaderived by mradermacher
    115.8 GiB
  • openai/gpt-oss-120bMXFP4vllmpublished by openai
    121.5 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
gpt-oss 120B
unsloth/gpt-oss-120b-GGUF:Q3_K_SQ3_K_Sllamacpp58.3 GiB on diskquantized by unsloth
unsloth/gpt-oss-120b-GGUF:Q2_KQ2_Kllamacpp58.3 GiB on diskquantized by unsloth
unsloth/gpt-oss-120b-GGUF:Q4_0Q4_0llamacpp58.3 GiB on diskquantized by unsloth
unsloth/gpt-oss-120b-GGUF:Q4_K_MQ4_K_Mllamacpp58.5 GiB on diskquantized by unsloth
mlx-community/gpt-oss-120b-MXFP4-Q84BITmlx59.0 GiB on diskquantized by mlx-community
unsloth/gpt-oss-120b-GGUF:Q8_0Q8_0llamacpp59.0 GiB on diskquantized by unsloth
hf.co/ggml-org/gpt-oss-120b-GGUF:MXFP4MXFP4ollama59.0 GiB on diskquantized by ggml-org
unsloth/gpt-oss-120b-GGUF:Q8_K_XLQ8_K_XLllamacpp60.0 GiB on diskquantized by unsloth
lmstudio-community/gpt-oss-120b-MLX-8bit8BITmlx115.6 GiB on diskquantized by lmstudio-community
openai/gpt-oss-120bMXFP4vllm121.5 GiB on diskpublished by openai
gpt-oss-120b-Fable-5-Distilled
cloudyu/gpt-oss-120b-Fable-5-Distilled4BITmlx59.0 GiB on diskderived by cloudyu
gpt-oss-120b-Eagle3-short-context
nvidia/gpt-oss-120b-Eagle3-short-contextBF16vllm1.9 GiB on diskderived by nvidia
gpt-oss-120b-abliterated
hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:MXFP4MXFP4ollama59.0 GiB on diskderived by mradermacher
hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q3_K_SQ3_K_Sollama61.6 GiB on diskderived by mradermacher
hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q2_KQ2_Kollama61.6 GiB on diskderived by mradermacher
hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q4_K_SQ4_K_Sollama75.4 GiB on diskderived by mradermacher
hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q4_K_MQ4_K_Mollama81.8 GiB on diskderived by mradermacher
hf.co/mradermacher/gpt-oss-120b-abliterated-GGUF:Q8_0Q8_0ollama115.8 GiB on diskderived by mradermacher
gpt-oss-tiny-random
yujiepan/gpt-oss-tiny-randomBF16vllm13 MiB on diskderived by yujiepan
gpt-oss 20B
hf.co/unsloth/gpt-oss-20b-GGUF:Q3_K_SQ3_K_Sollama10.7 GiB on diskquantized by unsloth
hf.co/unsloth/gpt-oss-20b-GGUF:Q2_KQ2_Kollama10.7 GiB on diskquantized by unsloth
hf.co/unsloth/gpt-oss-20b-GGUF:Q4_0Q4_0ollama10.7 GiB on diskquantized by unsloth
hf.co/unsloth/gpt-oss-20b-GGUF:Q4_K_MQ4_K_Mollama10.8 GiB on diskquantized by unsloth
mlx-community/gpt-oss-20b-MXFP4-Q84BITmlx11.2 GiB on diskquantized by mlx-community
hf.co/unsloth/gpt-oss-20b-GGUF:Q8_0Q8_0ollama11.3 GiB on diskquantized by unsloth
hf.co/unsloth/gpt-oss-20b-GGUF:Q8_K_XLQ8_K_XLollama12.3 GiB on diskquantized by unsloth
hf.co/unsloth/gpt-oss-20b-GGUF:F16F16ollama12.8 GiB on diskquantized by unsloth
lmstudio-community/gpt-oss-20b-MLX-8bit8BITmlx20.7 GiB on diskquantized by lmstudio-community
openai/gpt-oss-20bMXFP4vllm25.6 GiB on diskpublished by openai
gpt-oss-20b-Derestricted
hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q2_KQ2_Kollama11.2 GiB on diskderived by MaziyarPanahi
hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q3_K_MQ3_K_Mollama12.0 GiB on diskderived by MaziyarPanahi
hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q3_K_LQ3_K_Lollama12.4 GiB on diskderived by MaziyarPanahi
hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q4_K_MQ4_K_Mollama14.7 GiB on diskderived by MaziyarPanahi
hf.co/MaziyarPanahi/gpt-oss-20b-Derestricted-GGUF:Q5_K_MQ5_K_Mollama15.7 GiB on diskderived by MaziyarPanahi
openai_gpt-oss-20b-GGUF-MXFP4-Experimental
hf.co/bartowski/openai_gpt-oss-20b-GGUF-MXFP4-Experimental:MXFP4MXFP4ollama11.3 GiB on diskderived by bartowski
OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_SIQ1_Sollama11.2 GiB on diskderived by mradermacher
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ2_XXSIQ2_XXSollama11.2 GiB on diskderived by mradermacher
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:IQ1_MIQ1_Mollama11.2 GiB on diskderived by mradermacher
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q3_K_SQ3_K_Sollama11.2 GiB on diskderived by mradermacher
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_0Q4_0ollama11.3 GiB on diskderived by mradermacher
hf.co/mradermacher/OpenAI-gpt-oss-20B-Claude-4.5-Opus-Heretic-Uncensored-i1-GGUF:Q4_K_MQ4_K_Mollama14.7 GiB on diskderived by mradermacher
OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4
nightmedia/OpenAI-gpt-oss-20B-INSTRUCT-Heretic-Uncensored-MXFP4-q8-hi-mlx4BITmlx11.4 GiB on diskderived by nightmedia
gpt-oss-20b-MLX-6bit-gs32
halley-ai/gpt-oss-20b-MLX-6bit-gs326BITmlx17.1 GiB on diskderived by halley-ai
Huihui-gpt-oss-20b-BF16-abliterated
hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:IQ2_XSIQ2_XSollama10.7 GiB on diskderived by bartowski
hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:Q4_0Q4_0ollama10.7 GiB on diskderived by bartowski
hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:Q3_K_SQ3_K_Sollama10.8 GiB on diskderived by bartowski
hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:Q4_K_MQ4_K_Mollama10.9 GiB on diskderived by bartowski
hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:IQ4_NLIQ4_NLollama11.8 GiB on diskderived by DavidAU
hf.co/DavidAU/OpenAi-GPT-oss-20b-abliterated-uncensored-NEO-Imatrix-gguf:Q8_0Q8_0ollama20.6 GiB on diskderived by DavidAU
hf.co/bartowski/huihui-ai_Huihui-gpt-oss-20b-BF16-abliterated-GGUF:BF16BF16ollama39.0 GiB on diskderived by bartowski
Huihui-gpt-oss-20b-BF16-abliterated-v2
huihui-ai/Huihui-gpt-oss-20b-mxfp4-abliterated-v2MXFP4vllm12.8 GiB on diskderived by huihui-ai
gpt-oss-20b-tq3
manjunathshiva/gpt-oss-20b-tq33BITmlx9.3 GiB on diskderived by manjunathshiva
gpt-oss-20b-OptiQ-4bit
mlx-community/gpt-oss-20b-OptiQ-4bit4BITmlx10.8 GiB on diskderived by mlx-community
InternVL3_5-GPT-OSS-20B-A4B-Preview-HF
OpenGVLab/InternVL3_5-GPT-OSS-20B-A4B-Preview-HFBF16vllm39.5 GiB on diskderived by OpenGVLab
gpt-oss-20b-heretic
hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:IQ4_NLIQ4_NLollama11.8 GiB on diskderived by DavidAU
hf.co/DavidAU/OpenAi-GPT-oss-20b-HERETIC-uncensored-NEO-Imatrix-gguf:Q8_0Q8_0ollama20.6 GiB on diskderived by DavidAU
gpt-oss-20b-unsloth-bnb-4bit
unsloth/gpt-oss-20b-unsloth-bnb-4bitBITSANDBYTESvllm11.7 GiB on diskderived by unsloth
unsloth/gpt-oss-20b-BF16BF16vllm39.0 GiB on diskderived by unsloth
gpt-oss-120b-DFlash
z-lab/gpt-oss-120b-DFlashBF16vllm1.5 GiB on diskderived by z-lab
EAGLE3-gpt-oss-20b-bf16
zhuyksir/EAGLE3-gpt-oss-20b-bf16BF16vllm1.6 GiB on diskderived by zhuyksir
gpt-oss SAFEGUARD-120B
lmstudio-community/gpt-oss-safeguard-120b-MLX-MXFP44BITmlx57.9 GiB on diskquantized by lmstudio-community
openai/gpt-oss-safeguard-120bMXFP4vllm60.8 GiB on diskpublished by openai
gpt-oss SAFEGUARD-20B
lmstudio-community/gpt-oss-safeguard-20b-MLX-MXFP44BITmlx10.4 GiB on diskquantized by lmstudio-community
openai/gpt-oss-safeguard-20bMXFP4vllm12.8 GiB on diskpublished by openai
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 gpt-oss family (epn.family.gpt-oss), and can my gram run any of it? See /inferences/gpt-oss

open a gramx with this ask →epn.family.gpt-oss