GLM 4.7 Flash REAP 23B A3B — Hardware Requirements & GPU Compatibility
Chat
GLM 4.7 Flash REAP 23B A3B is a 23.0B-parameter open language model from Cerebras in the GLM 4 family. It supports a context window of up to 202,752 tokens. At Q4_K_M it needs about 14.89 GB of VRAM — see which GPUs and Macs can run it below.
Runs great on a 24 GB GPU
Runs from 8 GB (IQ2_XXS · 16K)Best on 24 GB (Q4_K_M · ~155 tok/s · 128K)Macs from 16 GB (Q3_K_M · ~41 tok/s)
MIT
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run GLM 4.7 Flash REAP 23B A3B
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
ollama run hf.co/unsloth/GLM-4.7-Flash-REAP-23B-A3B-GGUF:Q4_K_MQ4_K_M fits fully on the RTX 4090 · 9.0 GB free · 16K context
Download: GLM-4.7-Flash-REAP-23B-A3B-Q4_K_M.gguf14.11 GBunsloth
Other ways
each opens step-by-stepRun in cloud
Fits on RTX A5000 (24 GB) (9 GB headroom) · Q4_K_M
- Generation speed
- ~131 tok/s
- generation speed
- Cost per 1M output tokens
- $0.34
- per 1M output tokens
How Much VRAM Does GLM 4.7 Flash REAP 23B A3B Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 10.9 GB | 11.9 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 198K context · ~185 tok/s | |
| 11.2 GB | 12.1 GB | 3-bit small — 3-bit small quantization | Fits: 198K context · ~183 tok/s | |
| 12.3 GB | 13.3 GB | 3-bit medium — 3-bit medium quantization | Fits: 198K context · ~174 tok/s | |
| 12.6 GB | 13.6 GB | 4-bit · legacy — 4-bit legacy quantization | Fits: 198K context · ~171 tok/s | |
| 14.9 GB | 15.9 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 128K context · ~155 tok/s | |
| 17.5 GB | 18.5 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 128K context · ~141 tok/s | |
| 20.1 GB | 21.1 GB | 6-bit, very good — 6-bit quantization, very good quality | Fits: 64K context · ~129 tok/s | |
| 24.1 GB | 25.1 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Does not fit: needs 24.2 GB |
Selected Q4_K_M
4-bit medium quantization15.0GBat 16K context
- Weights
- 13.8 GB
- KV cache
- 0.9 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 9.0 GB free · ~155 tok/s
≥50 tok/s needs ≥223 GB/s · RTX 5090 ~211, RTX 3090 Ti ~155
Which Hardware Can Run GLM 4.7 Flash REAP 23B A3B?
Q4_K_M · 14.9 GBGLM 4.7 Flash REAP 23B A3B (Q4_K_M) requires 14.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 20+ GB is recommended. Using the full 203K context window can add up to 10.2 GB, bringing total usage to 25.1 GB. 26 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti, NVIDIA GeForce RTX 5080. 122 devices with unified memory can run GLM 4.7 Flash REAP 23B A3B, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Mini M5 Pro (24 GB).
16–18 GB
16–18 GB tier →fits · 1.0 GB free · up to 32K ctx
18 GPUs
RTX 5080~151Simulate output on RTX 5080RTX 5070 Ti~145Simulate output on RTX 5070 TiRTX 4080 SUPER~127Simulate output on RTX 4080 SUPER20 GB
fits · 5.0 GB free · up to 64K ctx
1 GPU
RX 7900 XT~68Simulate output on RX 7900 XT32–40 GB
fits · 17.0 GB free · up to 198K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~211Simulate output on RTX 5090M5 Max36 GB~58Simulate output on Mac Studio M5 Max (36 GB)M532 GB~30Simulate output on MacBook Air 13" M5 (32 GB)Orin~46Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~30Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
Alternatives to GLM 4.7 Flash REAP 23B A3B
GLM 4 family · 18 models →Similar models
4| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Magistral Small 2509Mistral AI | 85.6 | 18.1 GB | ~43 | same size |
| LFM2.5 24B A2BLiquid AI | — | 15 GB | ~221 | same size |
| EuroLLM 22B Instruct 2512utter-project | — | 17.5 GB | ~46 | same size |
| ERNIE 4.5 21B A3B ThinkingBaidu | — | 14.3 GB | ~213 | same size |
Specifications
- Publisher
- Cerebras
- Family
- GLM 4
- Parameters
- 23.0B
- Active parameters
- 3.9B per token
- Architecture
- Glm4MoeLiteForCausalLM
- Context Length
- 202,752 tokens
- Vocabulary Size
- 154,880
- Release Date
- 2026-01-23
- License
- MIT
Frequently Asked Questions
How much VRAM does GLM 4.7 Flash REAP 23B A3B need?
GLM 4.7 Flash REAP 23B A3B requires 14.9 GB of VRAM at Q4_K_M, or 47.1 GB at BF16. Full 203K context adds up to 10.2 GB (25.1 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 23.0B × 4.8 bits ÷ 8 = 13.8 GB
KV Cache + Overhead ≈ 1.1 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 ≈ 11.3 GB (at full 203K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 4090 run GLM 4.7 Flash REAP 23B A3B?
Yes, at Q6_K (20.1 GB) or lower. Higher quantizations like Q8_0 (24.1 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
What's the best quantization for GLM 4.7 Flash REAP 23B A3B?
For GLM 4.7 Flash REAP 23B A3B, Q4_K_M (14.9 GB) offers the best balance of quality and VRAM usage. Q5_K_S (16.9 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 7.4 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run GLM 4.7 Flash REAP 23B A3B on a Mac?
Yes — Mac Mini M2 Pro (16 GB) and 133 other Macs can run GLM 4.7 Flash REAP 23B A3B, the smallest at Q3_K_M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 14.9 GB of usable unified memory (RAM minus macOS overhead), from Mac Mini M5 Pro (24 GB).
Can I run GLM 4.7 Flash REAP 23B A3B locally?
Yes — GLM 4.7 Flash REAP 23B A3B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is GLM 4.7 Flash REAP 23B A3B?
At Q4_K_M, GLM 4.7 Flash REAP 23B A3B can reach ~101 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~155 tok/s. 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 → 2.5 GB active ÷ (8000 × 0.65) = 0.49 ms, plus 47 layers × 0.055 ms = 2.58 ms, so 1000 ÷ 3.07 ms = ~326 tok/s
Estimated speed at Q4_K_M (14.9 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of GLM 4.7 Flash REAP 23B A3B?
At Q4_K_M, the download is about 13.80 GB. The full-precision BF16 version is 45.99 GB. The smallest option (IQ2_XXS) is 6.32 GB.
Which GPUs can run GLM 4.7 Flash REAP 23B A3B?
26 consumer GPUs can run GLM 4.7 Flash REAP 23B A3B at Q4_K_M. Top options include AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090, NVIDIA GeForce RTX 3090 Ti, AMD Radeon RX 6800. 7 GPUs have plenty of headroom for comfortable inference.
Which devices can run GLM 4.7 Flash REAP 23B A3B?
122 devices with unified memory can run GLM 4.7 Flash REAP 23B A3B at Q4_K_M (14.9 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.