MiniMax M2.7 — Hardware Requirements & GPU Compatibility
ChatFunctions
MiniMax-M2.7 is MiniMax's latest agentic Mixture-of-Experts model, continuing the M2 line with roughly 229 billion total parameters. It's built for complex coding and agentic workflows, including multi-agent "Agent Teams" collaboration, long-horizon software engineering tasks like debugging and refactoring, and general productivity work such as editing documents and spreadsheets, with part of its own development driven by an internal self-improvement loop. It supports a 204,800-token context window and is released under a custom license. At this scale, 4-bit quantization needs roughly 132GB of memory, so local deployment realistically requires a multi-GPU server rather than a single workstation.
MiniMax M2.7 runs on a 192 GB+ Mac or a datacenter GPU
Best local pick → Mac Pro M2 Ultra (192 GB) · Q4_K_M · 139.6 GB · ~41 tok/s
Macs from 96 GB (IQ3_XXS · ~50 tok/s)
Score 107.4#34 of 100 open#17 of 78 that fit ≤192 GBOther
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
llmrun Composite Score →
107.4
- 90% range 102.0–112.7
- #34 of 100 open models
- #17 of 78 that fit ≤192 GB
Run MiniMax M2.7
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
ollama run hf.co/unsloth/MiniMax-M2.7-GGUF:Q4_K_MQ4_K_M runs on one H200 NVL 141 GB · 16K context
Download: UD-Q4_K_M/MiniMax-M2.7-UD-Q4_K_M-00001-of-00004.gguf140.17 GB + 3 partsunsloth
Other ways
each opens step-by-stepRun in cloud
Fits on B200 (192 GB) (52 GB headroom) · Q4_K_M
- Generation speed
- ~213 tok/s
- generation speed
- Cost per 1M output tokens
- $7.79
- per 1M output tokens
How Much VRAM Does MiniMax M2.7 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 97.8 GB | 101.7 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Runs on Instinct MI250X 128 GB | |
| 100.6 GB | 104.5 GB | 3-bit small — 3-bit small quantization | Runs on Instinct MI250X 128 GB | |
| 112.0 GB | 116 GB | 3-bit medium — 3-bit medium quantization | Runs on Instinct MI250X 128 GB | |
| 114.9 GB | 118.8 GB | 4-bit · legacy — 4-bit legacy quantization | Runs on Instinct MI250X 128 GB | |
| 137.8 GB | 141.7 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Runs on H200 NVL 141 GB | |
| 163.5 GB | 167.4 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Runs on B200 192 GB | |
| 189.2 GB | 193.1 GB | 6-bit, very good — 6-bit quantization, very good quality | Runs on B200 192 GB | |
| 229.3 GB | 233.2 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Runs on Instinct MI325X 256 GB |
Selected Q4_K_M
4-bit medium quantization139.6GBat 16K context
- Weights
- 137.2 GB
- KV cache
- 2.1 GB
- Overhead
- 0.3 GB
Server-class: ~139.6 GB at 16K context · one H200 NVL 141 GB
Which Hardware Can Run MiniMax M2.7?
Q4_K_M · 137.8 GBMiniMax M2.7 (Q4_K_M) requires 137.8 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 180+ GB is recommended. Using the full 205K context window can add up to 25.8 GB, bringing total usage to 163.5 GB. No consumer GPU has enough memory. 4 devices with unified memory can run MiniMax M2.7, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Pro M2 Ultra (192 GB).
128–144 GB
Q4_K_M · best fit IQ4_XS (125.3 GB)
4 GPUs · 2 Mac classes
GH200 Grace Hopper Superchip~189Simulate output on GH200 Grace Hopper SuperchipH200 NVL~188Simulate output on H200 NVLH200 SXM~188Simulate output on H200 SXMM2 Ultra*128 GB~43Simulate output on Mac Pro M2 Ultra (128 GB)M5 Max128 GB~38Simulate output on Mac Studio M5 Max (128 GB)192+ GB
fits · 52.4 GB free · up to 200K ctx
6 GPUs · 1 Mac class · 2 device families
B200~213Simulate output on B200B300~213Simulate output on B300Instinct MI350X~73Simulate output on Instinct MI350XM3 Ultra*256–512 GB~42Simulate output on Mac Studio (M3 Ultra, 256GB)8x H100 SXM5~264Simulate output on NVIDIA DGX H1008x A100 SXM4~248Simulate output on NVIDIA DGX A100 640GB
By VRAM tier · typical GPUs and Macs at Q4_K_M
Consumer GPUs (up to 32 GB) and Macs up to 64 GB: too small for this model (needs ≥ 65.3 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
How good is MiniMax M2.7?
The llmrun Score merges 7 public benchmark boards into a single figure, where 100 represents the average for open models.
Score vs VRAM in the ≤192 GB tier
- MiniMax M2.7
- Other models
- Best Score for its VRAM
#17 of 78 that fit ≤192 GB
Each dot is a model at the best quant that fits ≤192 GB. The line joins the models with the best Score for their VRAM; a dot below it is beaten by a model that needs no more memory.
Full ≤192 GB ranking →Benchmarks
The 6 boards that weigh most in the llmrun Score, of 9 with a result for this model; badges show who ran the test.
- LMArena Text
LMArena · Oct 2026Arena
Arena1415.0
#40/187 open
- Terminal-Bench
Epoch AI · Mar 2026Lab-reported
Lab-reported45.1%
#2/16 open
- LMArena WebDev
LMArena · Oct 2026Arena
Arena1398.7
#29/52 open
- SciCode
Epoch AI · Mar 2026Lab-reported
Lab-reported50.1%
#11/66 open
- WeirdML
Epoch AI · Mar 2026Independent
Independent37.0%
#30/46 open
- ALE-Bench
Epoch AI · Mar 2026Lab-reported
Lab-reported599.3
#27/43 open
In the ≤192 GB class
78 models fit ≤192 GBSame Score, only models that fit ≤192 GB at their best quant: the ranking a ≤192 GB owner actually chooses from.
- #1
MiMo V2.6 Flash RLXiaomi
131.9
~229 tok/s · 186.5 GB
- #2
DeepSeek V4 Flash 0731DeepSeek
127.6
~181 tok/s · 183.0 GB
- #3
GLM 5.3 FlashZ.ai
124.2
~228 tok/s · 185.1 GB
- #4
Qwen3.8 27BAlibaba
122.1
~299 tok/s · 17.9 GB
- #5
Inkling SmallThinking Machines
119.6
~399 tok/s · 162.7 GB
- #6
Qwen3.5 397B A17BAlibaba
113.0
~185 tok/s · 182.1 GB
Alternatives to MiniMax M2.7
MiniMax family · 9 models →Official models with an llmrun Score, each answering one question; sizes are each model's default quant (usually Q4_K_M) with a 16K-token context.
Best alternative at this size
Inkling Small
Thinking Machines
- 12.2 higher Score
- 23.1 GB more VRAM
Score 119.6 · 162.7 GB
Other vendor
MiMo V2.6 Flash RL
Xiaomi
- 24.5 higher Score
- 46.9 GB more VRAM
Score 131.9 · 186.5 GB
Similar models
10| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| DeepSeek V4 Flash 0731DeepSeek | 127.6 | 183 GB | —needs > 24 GB | similar score |
| GLM 5.3 FlashZ.ai | 124.2 | 205.2 GB | —needs > 24 GB | similar score |
| Qwen3.5 397B A17BAlibaba | 113.0 | 242.6 GB | —needs > 24 GB | similar score |
| MiniMax M3MiniMax | 112.5 | 258 GB | —needs > 24 GB | similar score |
| MiniMax M2.5MiniMax | 111.8 | 139.6 GB | —needs > 24 GB | similar score |
| Qwen3.5 122B A10BAlibaba | 109.4 | 75.5 GB | —needs > 24 GB | similar score |
| DeepSeek V4 FlashDeepSeek | 109.3 | 175.1 GB | —needs > 24 GB | similar score |
| GLM 4.7Z.ai | 109.1 | 217.9 GB | —needs > 24 GB | similar score |
| MiMo V2.5Xiaomi | 108.3 | 187.6 GB | —needs > 24 GB | similar score |
| Gemma 4 31B ITGoogle | 106.8 | 21.7 GB | ~32 | similar score |
Fine-tunes & variants(2)
Specifications
- Publisher
- MiniMax
- Family
- MiniMax
- Parameters
- 228.7B
- Active parameters
- 11B per token
- Architecture
- MiniMaxM2ForCausalLM
- Context Length
- 204,800 tokens
- Vocabulary Size
- 200,064
- Release Date
- 2026-04-09
- License
- Other
Frequently Asked Questions
How much VRAM does MiniMax M2.7 need?
MiniMax M2.7 requires 137.8 GB of VRAM at Q4_K_M, or 457.9 GB at BF16. Full 205K context adds up to 25.8 GB (163.5 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 228.7B × 4.8 bits ÷ 8 = 137.2 GB
KV Cache + Overhead ≈ 0.6 GB (at 2K context + ~0.3 GB framework)
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
KV Cache + Overhead ≈ 26.3 GB (at full 205K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run MiniMax M2.7?
No — MiniMax M2.7 requires at least 63.5 GB at IQ2_XXS, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
What's the best quantization for MiniMax M2.7?
For MiniMax M2.7, Q4_K_M (137.8 GB) offers the best balance of quality and VRAM usage. Q4_K_L (140.6 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 63.5 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run MiniMax M2.7 on a Mac?
Yes — Mac Studio (M3 Ultra, 96GB) and 20 other Macs can run MiniMax M2.7, the smallest at IQ3_XXS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 137.8 GB of usable unified memory (RAM minus macOS overhead), from Mac Pro M2 Ultra (192 GB).
Can I run MiniMax M2.7 locally?
Yes — MiniMax M2.7 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is MiniMax M2.7?
At Q4_K_M, MiniMax M2.7 can reach ~73 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)
Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.
Example: NVIDIA B200 → 6.6 GB active ÷ (8000 × 0.65) = 1.28 ms, plus 62 layers × 0.055 ms = 3.41 ms, so 1000 ÷ 4.69 ms = ~213 tok/s
Estimated speed at Q4_K_M (137.8 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of MiniMax M2.7?
At Q4_K_M, the download is about 137.21 GB. The full-precision BF16 version is 457.38 GB. The smallest option (IQ2_XXS) is 62.89 GB.
Which GPUs can run MiniMax M2.7?
No single consumer GPU has enough VRAM to run MiniMax M2.7 at Q4_K_M (137.8 GB). Multi-GPU or professional hardware is required.
Which devices can run MiniMax M2.7?
4 devices with unified memory can run MiniMax M2.7 at Q4_K_M (137.8 GB), including Mac Pro M2 Ultra (192 GB), Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Studio M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.