GLM 4.6 — Hardware Requirements & GPU Compatibility
ChatFunctions
GLM-4.6 is Z.ai's (Zhipu AI's) flagship chat and agentic model, a mixture-of-experts system with about 356.8 billion total parameters and roughly 34 billion active per token. Compared with the prior GLM-4.5, it extends the context window, improves coding performance in real-world tools such as Claude Code and Cline, strengthens tool-use and search-agent capability during inference, and refines writing style and role-play naturalism; Z.ai reports it is competitive with models like DeepSeek-V3.1-Terminus and Claude Sonnet 4 on agent, reasoning, and coding benchmarks. It shares its inference method and underlying architecture with GLM-4.5. Given its total parameter count, it needs a multi-GPU server-class setup even once quantized. Context length is 200,000 tokens, up from 128K in GLM-4.5. It is released under the MIT license, permitting unrestricted commercial and research use. It was published in September 2025.
GLM 4.6 runs on a 256 GB+ Mac or a datacenter GPU
Best local pick → Mac Studio (M3 Ultra, 256GB) · Q4_K_M · 216.9 GB · ~19 tok/s
Macs from 128 GB (IQ2_M · ~26 tok/s)
Score 103.1#45 of 100 open#32 of 84 that fit ≤256 GBMIT
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
llmrun Composite Score →
103.1
- 90% range 98.4–107.9
- #45 of 100 open models
- #32 of 84 that fit ≤256 GB
Run GLM 4.6
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
ollama run hf.co/unsloth/GLM-4.6-GGUF:Q4_K_MQ4_K_M runs on one Instinct MI325X 256 GB · 16K context
Download: Q4_K_M/GLM-4.6-Q4_K_M-00001-of-00005.gguf215.58 GB + 4 partsunsloth
Other ways
each opens step-by-stepRun in cloud
Fits on B300 (288 GB) (71 GB headroom) · Q4_K_M
- Generation speed
- ~111 tok/s
- generation speed
- Cost per 1M output tokens
- $17.35
- per 1M output tokens
How Much VRAM Does GLM 4.6 Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 152.3 GB | 157.1 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Runs on B200 192 GB | |
| 156.7 GB | 161.5 GB | 3-bit small — 3-bit small quantization | Runs on B200 192 GB | |
| 174.6 GB | 179.4 GB | 3-bit medium — 3-bit medium quantization | Runs on B200 192 GB | |
| 179.0 GB | 183.8 GB | 4-bit · legacy — 4-bit legacy quantization | Runs on B200 192 GB | |
| 214.7 GB | 219.5 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Runs on Instinct MI325X 256 GB | |
| 254.8 GB | 259.7 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Runs on Instinct MI350X 288 GB | |
| 295.0 GB | 299.8 GB | 6-bit, very good — 6-bit quantization, very good quality | Runs on 5× H100 SXM 80 GB | |
| 357.4 GB | 362.2 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Runs on 5× H100 SXM 80 GB |
Selected Q4_K_M
4-bit medium quantization216.9GBat 16K context
- Weights
- 214.1 GB
- KV cache
- 2.6 GB
- Overhead
- 0.3 GB
Server-class: ~216.9 GB at 16K context · one Instinct MI325X 256 GB
Which Hardware Can Run GLM 4.6?
Q4_K_M · 214.7 GBGLM 4.6 (Q4_K_M) requires 214.7 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 280+ GB is recommended. Using the full 203K context window can add up to 31.5 GB, bringing total usage to 246.2 GB. No consumer GPU has enough memory. 2 devices with unified memory can run GLM 4.6, including Mac Studio (M3 Ultra, 512GB), Mac Studio (M3 Ultra, 256GB).
128–144 GB
Q4_K_M · best fit IQ2_M (123.3 GB)
4 GPUs · 2 Mac classes
GH200 Grace Hopper Superchip~115Simulate output on GH200 Grace Hopper SuperchipH200 NVL~114Simulate output on H200 NVLH200 SXM~114Simulate output on H200 SXMM2 Ultra*128 GB~26Simulate output on Mac Pro M2 Ultra (128 GB)M5 Max128 GB~22Simulate output on Mac Studio M5 Max (128 GB)192+ GB
Q4_K_M · best fit Q3_K_L (185.7 GB)
6 GPUs · 1 Mac class · 2 device families
B200~119Simulate output on B200B300~119Simulate output on B300Instinct MI350X~45Simulate output on Instinct MI350XM3 Ultra*256–512 GB~20Simulate output on Mac Studio (M3 Ultra, 256GB)8x H100 SXM5~165Simulate output on NVIDIA DGX H1008x A100 SXM4~149Simulate 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 96 GB: too small for this model (needs ≥ 101 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
How good is GLM 4.6?
The llmrun Score merges 6 public benchmark boards into a single figure, where 100 represents the average for open models.
Score vs VRAM in the ≤256 GB tier
- GLM 4.6
- Other models
- Best Score for its VRAM
#32 of 84 that fit ≤256 GB
Each dot is a model at the best quant that fits ≤256 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 ≤256 GB ranking →Benchmarks
The 6 boards that weigh most in the llmrun Score, of 8 with a result for this model; badges show who ran the test.
- SWE-bench Verified
SWE-bench · Sep 2025Independent
Independent68.2%
#7/20 open
- LMArena Text
LMArena · Oct 2026Arena
Arena1424.1
#31/187 open
- Terminal-Bench
Epoch AI · Sep 2025Lab-reported
Lab-reported24.5%
#13/16 open
- LMArena WebDev
LMArena · Oct 2026Arena
Arena1341.1
#40/52 open
- SWE-bench Bash Only
SWE-bench · Dec 2025Independent
Independent55.4%
#5/9 open
- SciCode
Epoch AI · Sep 2025Lab-reported
Lab-reported38.4%
#35/66 open
In the ≤256 GB class
84 models fit ≤256 GBSame Score, only models that fit ≤256 GB at their best quant: the ranking a ≤256 GB owner actually chooses from.
- #1
MiMo V2.6 Flash RLXiaomi
131.9
~83 tok/s · 186.5 GB
- #2
DeepSeek V4.1 FlashDeepSeek
129.5
~45 tok/s · 210.4 GB
- #3
DeepSeek V4 Flash 0731DeepSeek
127.6
~76 tok/s · 183.0 GB
- #4
GLM 5.3 FlashZ.ai
124.2
~83 tok/s · 205.2 GB
- #5
GLM 5.2Z.ai
122.9
~52 tok/s · 227.8 GB
- #6
Qwen3.8 27BAlibaba
122.1
~207 tok/s · 17.9 GB
Alternatives to GLM 4.6
GLM 4 family · 18 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
MiMo V2.6 Flash RL
Xiaomi
- 28.8 higher Score
- 30.4 GB less VRAM
Score 131.9 · 186.5 GB
Fits in less VRAM
Inkling Small
Thinking Machines
- 16.5 higher Score
- 54.2 GB less VRAM
Score 119.6 · 162.7 GB
Other vendor
DeepSeek V4 Flash 0731
DeepSeek
- 24.5 higher Score
- 33.9 GB less VRAM
Score 127.6 · 183 GB
Similar models
10| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| 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 |
| DeepSeek V3.2DeepSeek | 110.4 | 412.7 GB | —needs > 24 GB | similar score |
| NVIDIA Nemotron 3 Ultra 550B A55B BF16NVIDIA | 109.9 | 378.4 GB | —needs > 24 GB | similar score |
| Qwen3 235B A22B Thinking 2507Alibaba | 108.2 | 142.9 GB | —needs > 24 GB | similar score |
| MiniMax M2.7MiniMax | 107.4 | 139.6 GB | —needs > 24 GB | similar score |
| DeepSeek V3.1 TerminusDeepSeek | 107.3 | 412.2 GB | —needs > 24 GB | similar score |
| MiniMax M2.1MiniMax | 106.2 | 139.6 GB | —needs > 24 GB | similar score |
Fine-tunes & variants(2)
Specifications
- Publisher
- Z.ai
- Family
- GLM 4
- Parameters
- 356.8B
- Active parameters
- 34B per token
- Architecture
- Glm4MoeForCausalLM
- Context Length
- 202,752 tokens
- Vocabulary Size
- 151,552
- Release Date
- 2025-09-29
- License
- MIT
Frequently Asked Questions
How much VRAM does GLM 4.6 need?
GLM 4.6 requires 214.7 GB of VRAM at Q4_K_M, or 714.2 GB at BF16. Full 203K context adds up to 31.5 GB (246.2 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 356.8B × 4.8 bits ÷ 8 = 214.1 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 ≈ 32.1 GB (at full 203K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run GLM 4.6?
No — GLM 4.6 requires at least 98.7 GB at IQ2_XXS, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
What's the best quantization for GLM 4.6?
For GLM 4.6, Q4_K_M (214.7 GB) offers the best balance of quality and VRAM usage. Q5_K_S (245.9 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 98.7 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run GLM 4.6 on a Mac?
Yes — Mac Pro M2 Ultra (128 GB) and 14 other Macs can run GLM 4.6, the smallest at IQ2_M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 214.7 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio (M3 Ultra, 256GB).
Can I run GLM 4.6 locally?
Yes — GLM 4.6 can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is GLM 4.6?
At Q4_K_M, GLM 4.6 can reach ~44 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 B300 → 20.5 GB active ÷ (8000 × 0.65) = 3.94 ms, plus 92 layers × 0.055 ms = 5.06 ms, so 1000 ÷ 9.00 ms = ~111 tok/s
Estimated speed at Q4_K_M (214.7 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of GLM 4.6?
At Q4_K_M, the download is about 214.07 GB. The full-precision BF16 version is 713.57 GB. The smallest option (IQ2_XXS) is 98.12 GB.
Which GPUs can run GLM 4.6?
No single consumer GPU has enough VRAM to run GLM 4.6 at Q4_K_M (214.7 GB). Multi-GPU or professional hardware is required.
Which devices can run GLM 4.6?
2 devices with unified memory can run GLM 4.6 at Q4_K_M (214.7 GB), including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.