- Size
- 427B parameterscounted from the safetensors header on the Hub
- Licence
- custom, read it, MiniMax Community LicenseOur reading, check the licence: A custom licence we have not read for this model: other makers use the same name for different texts. Read the model's own licence file before you use it.Licence text
- On the Hub since
- 2 Jun 2026the date the repo was created
- Context length
- 1M tokensfrom the model's config
- Kind
- Base model (no base model declared on the Hub)MiniMaxM3SparseForConditionalGeneration
- On Hugging Face
- MiniMaxAI/MiniMax-M3104k downloads in the last 30 days, 1.6k likes
Scores
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5-2_max, 79.0% (Anthropic). All 166 results
Epoch's name: MiniMax-M3. Best closed model on this set: gpt-6.2-astra_max, 96.8% (OpenAI). All 341 results
Epoch's name: MiniMax-M3. Best closed model on this set: gpt-6.2-astra_max, 95.0% (OpenAI). All 148 results
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5_high, 100.0% (Anthropic). All 318 results
- default setting37.7%via Epoch AI (error bar left out)
Epoch's name: MiniMax-M3. Best closed model on this set: gemini-4-carbon_unknown, 84.2% (Google DeepMind). All 66 results
- default setting3.7%via Epoch AI
Epoch's name: MiniMax-M3. Best closed model on this set: gemini-4-carbon_high, 36.0% (Google DeepMind). All 271 results
- default setting78.9%conceptualreasoning.ai
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5-2_unknown, 99.2% (Anthropic). All 241 results
- Harness: mini-swe-agent; Reasoning effort: none14.7%cognition.com
Epoch's name: MiniMax-M3. Best closed model on this set: claude-opus-5-5_medium, 54.6% (Anthropic). All 46 results
- default setting0.9%2 Jun 2026, gbaeval.com
Epoch's name: MiniMax-M3. Best closed model on this set: claude-opus-5-5_unknown, 90.6% (Anthropic). All 28 results
- default setting33.7%29.2 to 38.2conceptualreasoning.ai
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5-2_xhigh, 70.6% (Anthropic). All 203 results
- Reasoning: enabled; Tool setting: standard; Step budget: 5004.6%osworld-v2.xlang.ai
Epoch's name: MiniMax-M3. Best closed model on this set: claude-opus-5_max, 31.4% (Anthropic). All 16 results
- default setting18.0%±3.9vals.ai
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5-1_max, 100.0% (Anthropic). All 85 results
- default setting45.8%lmcouncil.ai
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5-2_unknown, 94.9% (Anthropic). All 125 results
- default setting55.0%yhenon.github.io
Epoch's name: MiniMax-M3. Best closed model on this set: claude-fable-5_high, 95.0% (Anthropic). All 29 results
Claimed by the maker Claimed
The model card has no claimed scores in a machine-readable form (the card's model-index), so none are shown. The maker may give scores in the card text: read the card.
Other views, which we link to and do not copy: look this model up on Artificial Analysis and the Arena leaderboard.
Independent scores from Epoch AI (CC-BY 4.0), fetched 19 Nov 2026: Epoch AI, ‘Capabilities & benchmarking’. Published online at epoch.ai. Retrieved from ‘https://epoch.ai/benchmarks’ [online resource]. Each row links its own source. What the badges mean.
Run it yourself
Memory needed = weights + context cache + 0.50 GB for the runtime. Context cache at 8k tokens: 0.94 GB, worked out from the model's config (MiniMaxAI/MiniMax-M3), at 16-bit precision. Choose a machine above to see a verdict for each format. Method
GGUF for llama.cpp, Ollama, LM Studio and others
unsloth/MiniMax-M3-GGUF (converted by unsloth), 12k downloads. Sizes are the real files on the Hub, all parts added up.
| Format | Size | Memory needed | Get it |
|---|---|---|---|
| UD-IQ3_XXS | 149.3 GBfile size, 5 parts | 150.7 GB149.3 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-IQ3_XXSDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-IQ3_XXS/MiniMax-M3-UD-IQ3_XXS-*-of-00005.gguf" --local-dir MiniMax-M3-GGUF |
| UD-IQ3_S | 163.6 GBfile size, 5 parts | 165.0 GB163.6 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-IQ3_SDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-IQ3_S/MiniMax-M3-UD-IQ3_S-*-of-00005.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q3_K_M | 182.1 GBfile size, 5 parts | 183.5 GB182.1 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q3_K_MDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q3_K_M/MiniMax-M3-UD-Q3_K_M-*-of-00005.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q3_K_XL | 182.3 GBfile size, 5 parts | 183.8 GB182.3 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q3_K_XLDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q3_K_XL/MiniMax-M3-UD-Q3_K_XL-*-of-00005.gguf" --local-dir MiniMax-M3-GGUF |
| UD-IQ4_XS | 194.1 GBfile size, 6 parts | 195.6 GB194.1 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-IQ4_XSDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-IQ4_XS/MiniMax-M3-UD-IQ4_XS-*-of-00006.gguf" --local-dir MiniMax-M3-GGUF |
| UD-IQ4_NL | 198.1 GBfile size, 6 parts | 199.5 GB198.1 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-IQ4_NLDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-IQ4_NL/MiniMax-M3-UD-IQ4_NL-*-of-00006.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q4_K_S | 231.5 GBfile size, 7 parts | 233.0 GB231.5 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q4_K_SDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q4_K_S/MiniMax-M3-UD-Q4_K_S-*-of-00007.gguf" --local-dir MiniMax-M3-GGUF |
| MXFP4_MOE | 239.7 GBfile size, 7 parts | 241.1 GB239.7 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:MXFP4_MOEDownload hf download unsloth/MiniMax-M3-GGUF --include "MXFP4_MOE/MiniMax-M3-MXFP4_MOE-*-of-00007.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q4_K_M | 246.7 GBfile size, 7 parts | 248.2 GB246.7 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q4_K_MDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q4_K_M/MiniMax-M3-UD-Q4_K_M-*-of-00007.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q4_K_XL | 247.6 GBfile size, 7 parts | 249.0 GB247.6 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q4_K_XLDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q4_K_XL/MiniMax-M3-UD-Q4_K_XL-*-of-00007.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q5_K_S | 279.1 GBfile size, 8 parts | 280.5 GB279.1 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q5_K_SDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q5_K_S/MiniMax-M3-UD-Q5_K_S-*-of-00008.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q5_K_M | 296.8 GBfile size, 8 parts | 298.3 GB296.8 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q5_K_MDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q5_K_M/MiniMax-M3-UD-Q5_K_M-*-of-00008.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q5_K_XL | 297.4 GBfile size, 8 parts | 298.9 GB297.4 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q5_K_XLDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q5_K_XL/MiniMax-M3-UD-Q5_K_XL-*-of-00008.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q6_K | 330.9 GBfile size, 9 parts | 332.3 GB330.9 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q6_KDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q6_K/MiniMax-M3-UD-Q6_K-*-of-00009.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q6_K_XL | 361.4 GBfile size, 9 parts | 362.9 GB361.4 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q6_K_XLDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q6_K_XL/MiniMax-M3-UD-Q6_K_XL-*-of-00009.gguf" --local-dir MiniMax-M3-GGUF |
| Q8_0 | 422.5 GBfile size, 11 parts | 423.9 GB422.5 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:Q8_0Download hf download unsloth/MiniMax-M3-GGUF --include "Q8_0/MiniMax-M3-Q8_0-*-of-00011.gguf" --local-dir MiniMax-M3-GGUF |
| UD-Q8_K_XL | 433.3 GBfile size, 11 parts | 434.7 GB433.3 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:UD-Q8_K_XLDownload hf download unsloth/MiniMax-M3-GGUF --include "UD-Q8_K_XL/MiniMax-M3-UD-Q8_K_XL-*-of-00011.gguf" --local-dir MiniMax-M3-GGUF |
| BF16 | 794.3 GBfile size, 18 parts | 795.8 GB794.3 + 0.94 + 0.50 | Commandsllama.cpp llama-server -hf unsloth/MiniMax-M3-GGUF:BF16Download hf download unsloth/MiniMax-M3-GGUF --include "BF16/MiniMax-M3-BF16-*-of-00018.gguf" --local-dir MiniMax-M3-GGUF |
MLX for Apple silicon
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| MLX 4bitpipenetwork/MiniMax-M3-MLX-4bit | 223.2 GBfile size | 224.6 GB223.2 + 0.94 + 0.50 | Download page |
| MLX 4bitmlx-community/MiniMax-M3-4bit | 224.9 GBfile size | 226.3 GB224.9 + 0.94 + 0.50 | Download page |
Original weights size from the safetensors header, for vLLM, SGLang, Transformers
| Format | Size | Memory needed | Repo |
|---|---|---|---|
| BF16original weights | 795.5 GBfile size | 796.9 GB795.5 + 0.94 + 0.50 |
hf download MiniMaxAI/MiniMax-M3 --local-dir MiniMax-M3Other conversions (15; sizes not read)
- MiniMaxAI/MiniMax-M3-MXFP8 Other, by the maker, 39k downloads
- nvidia/MiniMax-M3-NVFP4 NVFP4, 38k downloads
- amd/MiniMax-M3-MXFP4-AttnFP8 Other, mxfp4, 27k downloads
- amd/MiniMax-M3-MXFP4 Other, mxfp4, 16k downloads
- Sebesky/MiniMax-M3-W4A16-GPTQ GPTQ, w4a16, 2.6k downloads
- lukealonso/MiniMax-M3-NVFP4 NVFP4, 362 downloads
- Kanposer/MiniMax-M3-speedy-colibri-nvfp4 NVFP4, 189 downloads
- intel-ai/MiniMax-M3-MXFP4-CT-AutoRound Other, mxfp4, 184 downloads
- cyankiwi/MiniMax-M3-AWQ-INT4 AWQ, int4, 171 downloads
- Arsh9210/MiniMax-M3-NVFP4 NVFP4, 162 downloads
- EmbeddedLLM/MiniMax-M3-FP8-dynamic FP8, 152 downloads
- spectator2026/MiniMax-M3-AWQ-int4 AWQ, int4, 132 downloads
- Model-SafeTensors/MiniMax-M3-MoE-FP8-128x128 FP8, 40 downloads
- aquaman164/MiniMax-M3-AutoRound-3.2bit-longctx GPTQ, 2bit, 35 downloads
- FenomAI/MiniMax-M3-AWQ-INT4 AWQ, int4, 22 downloads
The Hub tags 56 quantised repos as conversions of this model; we read the most downloaded and show the maker's own first.
Run it by API
6 providers serve this model through the Hugging Face router. Prices are US dollars per million tokens. Precision is what the provider reports; a lower precision than the original weights can change answers. Live first, then cheapest output.
| Provider | In | Out | Context | Precision | Tokens/s | First token | Uptime 24h | Tools | JSON |
|---|---|---|---|---|---|---|---|---|---|
| CoreWeave | $0.23 | $0.96 | 256k | fp4 | 102 | 0.6 s | 94.2% | yes | yes |
| DeepInfra | $0.28 | $1.10 | 512k | fp8 | 66 | 0.7 s | 98.9% | yes | no |
| novita | $0.30 | $1.20 | 977k | not given | 56 | 0.8 s | ? | yes | no |
| Parasail | $0.30 | $1.20 | 1M | fp8 | 85 | 0.6 s | 99.9% | yes | no |
| Venice | $0.30 | $1.20 | 512k | fp8 | 134 | 0.8 s | 97.6% | yes | no |
| Featherless | $0.55 | $2.20 | 256k | fp8 | 96 | 0.9 s | 10.3% | no | no |
Through the router, with a Hugging Face token (the router bills your Hugging Face account). Change the name after the colon to pick another provider.
curl https://router.huggingface.co/v1/chat/completions \
-H "Authorization: Bearer $HF_TOKEN" -H "Content-Type: application/json" \
-d '{"model": "MiniMaxAI/MiniMax-M3:coreweave", "messages": [{"role": "user", "content": "Hello"}]}'Not built yet: one key for every open model, with BlockRun, and our own tested-quality score per provider. Until then, precision and measured speed are the best guide to a provider here.
Fine-tunes of this model
Repos that declare this model as their base, most downloaded first. Listing is not a recommendation; unreviewed ones have not been checked.
- nvidia/MiniMax-M3-DSpark3.3B1.3kUnreviewed
Planned for this page Not built yet
- Try it in private chat
- A link to Solana AI's confidential chat for the models it serves. This model is not listed as served there.
- Fine-tune it
- Fine-tuning a few base models from this page and publishing the result here.
- Creators and earnings
- Pages for people who publish fine-tunes and adapters of this model, and what they earn.
- Fund and bounties
- Bounties for new open models, paid from a public fund. How the fund will work.
- Contribute compute
- How to give compute to open model work, with Akash.
Model data from the Hugging Face Hub 19 Nov 2026, 13:43 UTC. The repo. Found a mistake on this page? Tell us.