RTX 3090 24GBコスパ重視24 GB VRAM (+4)936 GB/s (+576)
C
Raises estimated decode speed by about 85%.109.2 tok/sデコード
Raises estimated decode speed by about 85%.
〜$1,499 MSRP
Can EXAONE 3.5 7.8B Instruct i1 run on RTX 4000 Ada 20GB?
YES — Runs Great
C50Usable
Estimated from fit model
EXAONE 3.5 7.8B Instruct i1 needs ~8.9 GB VRAM. RTX 4000 Ada 20GB has 20.0 GB. With Q4_K_M quantization, expect ~59 tok/s.
Runtime: OllamaCapacity: RoomyBandwidth: LowStack: BasicBottleneck: Balanced
Operating mode
Choose the run profile you care about
Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.
Current mode
Balanced
Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.
Select quantization to explore
Q4_K_M (Medium quality) — 8.9 GB, 59.0 tok/s, Runs well
8.9 GB required20.0 GB available
45% VRAM used
Fit status
Runs well
Decode
59.0 tok/s
TTFT
3280 ms
Safe context
211K
Memory
8.9 GB / 20.0 GB
Memory breakdown
Weights4.8 GB
KV Cache0.9 GB
Runtime1.2 GB
Headroom2.0 GB
See how fast it feels
See how fast it feelsEXAONE 3.5 7.8B Instruct i1 on RTX 4000 Ada 20GB
1st promptCold start — includes initialization
>What is local AI inference?
Local AI inference means running an AI language model directly on your own hardware — your laptop, desktop, or server — instead of sending requests to a remote cloud API.
When you run inference locally the model weights are loaded into your GPU or unified memory. Each token you generate requires reading those weights from memory, so memory bandwidth is the main bottleneck for decode speed.
Key benefits of running locally:
- Full privacy: your prompts never leave your machine
- No per-token cost or rate limits
- Works offline once the model is downloaded
- Latency depends only on your hardware
2nd promptWarm — model ready, no init overhead
>How much VRAM do I need?
It depends on the model size and quantization level. A rough rule of thumb:
Model size Q4 (4-bit) Q8 (8-bit) FP16
7B params ~4.3 GB ~7.5 GB ~14 GB
13B params ~7.9 GB ~13.9 GB ~26 GB
70B params ~42.7 GB ~74.9 GB ~140 GB
Most people use 4-bit quantization (Q4_K_M) which gives 90-95% of full quality at a fraction of the memory. A 24 GB GPU can comfortably run most 7B-13B models.
Estimated: 59.0 tok/s decode · 3.3s TTFT (warm) · 148 tok/s prefill
What limits this setup
This setup is broadly balanced for this model.
No major red flags
This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.
Best improvement path
Performance by workload
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | C | Runs well | 59.0 tok/s | 1789 ms | 211K |
| Coding | C | Runs well | 59.0 tok/s | 3280 ms | 211K |
| Agentic Coding | C | Runs well | 59.0 tok/s | 4772 ms | 211K |
| Reasoning | C | Runs well | 59.0 tok/s | 3877 ms | 211K |
| RAG | C | Runs well | 59.0 tok/s | 5964 ms | 211K |
Inference speed
EXAONE 3.5 7.8B Instruct i1 inference speed — tokens per second by GPU & Mac
Estimated decode speed (tokens/sec) for EXAONE 3.5 7.8B Instruct i1 at Q4_K_M across popular GPUs and Apple Silicon, using the fastest local runtime per device. Fastest is RTX 5090 32GB at ~109 tok/s. Speed is memory-bandwidth bound, so cards that fit the whole model in VRAM run far faster than ones that offload to system RAM.
| GPU / Mac | Memory | Quant | Speed (tok/s) | Fits? |
|---|---|---|---|---|
| RTX 5090 32GB | 32 GB | Q4_K_M | 109.2 | Fits |
| RTX 4090 24GB | 24 GB | Q4_K_M | 109.2 | Fits |
| RTX 4080 Super 16GB | 16 GB | Q4_K_M | 109.2 | Fits |
| RTX 3090 24GB | 24 GB | Q4_K_M | 109.2 | Fits |
RX 7900 XTX 24GB | 24 GB | Q4_K_M | 109.2 | Fits |
Mac Studio M3 Ultra 256GB | 256 GB | Q4_K_M | 109.2 | Fits |
Mac Studio M2 Ultra 128GB | 128 GB | Q4_K_M | 97.5 | Fits |
Mac Studio M1 Ultra 128GB | 128 GB | Q4_K_M | 92.5 | Fits |
| RTX 4070 12GB | 12 GB | Q4_K_M | 79.4 | Fits |
MacBook Pro M4 Max 128GB | 128 GB | Q4_K_M | 78.8 | Fits |
MacBook Pro M4 Max 64GB | 64 GB | Q4_K_M | 78.8 | Fits |
MacBook Pro M3 Max 64GB | 64 GB | Q4_K_M | 50.4 | Fits |
| RTX 3060 12GB | 12 GB | Q4_K_M | 49.9 | Fits |
MacBook Pro M1 Max 64GB | 64 GB | Q4_K_M | 46.2 | Fits |
| RTX 4060 8GB | 8 GB | Q4_K_M | 41.7 | Offloads |
MacBook Pro M4 Pro 48GB | 48 GB | Q4_K_M | 40.6 | Fits |
Estimates for single-stream decoding at Q4_K_M; real tokens/sec varies with prompt length, context, batch size, and runtime build. Prompt processing (prefill) is faster than the decode figures shown here.
Quantization options
How EXAONE 3.5 7.8B Instruct i1 (7.800000190734863B params) fits at each quantization level on RTX 4000 Ada 20GB (20.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 3.0 GB | Low | C45 |
Q3_K_S | 3 | 3.8 GB | Low | C46 |
NVFP4 | 4 | 4.4 GB | Medium | C46 |
Q4_K_M | 4 | 4.8 GB | Medium | C46 |
Q5_K_M | 5 | 5.6 GB | High | C47 |
Q6_K | 6 | 6.4 GB | High | C48 |
Q8_0 | 8 | 8.3 GB | Very High | C49 |
F16Best for your GPU | 16 | 16.0 GB | Maximum | C50 |
Get started
Copy-paste commands to run EXAONE 3.5 7.8B Instruct i1 on your machine.
Run
lms load hf-mradermacher--exaone-3-5-7-8b-instruct-i1-gguf && lms server startアップグレードオプション
EXAONE 3.5 7.8B Instruct i1を快適に動かすハードウェア
RTX 3090 24GBコスパ重視RTX 4090 24GBコスパ最良24 GB VRAM (+4)1008 GB/s (+648)
C
Raises estimated decode speed by about 85%.109.2 tok/sデコード
Raises estimated decode speed by about 85%.
〜$1,599 MSRP
RTX PRO 4000 Blackwell 24GBNVIDIAアップグレード24 GB VRAM (+4)672 GB/s (+312)
C
Raises estimated decode speed by about 85%.109.2 tok/sデコード
Raises estimated decode speed by about 85%.
〜$1,599 MSRP
Frequently asked questions
Can RTX 4000 Ada 20GB run EXAONE 3.5 7.8B Instruct i1?
Yes, RTX 4000 Ada 20GB can run EXAONE 3.5 7.8B Instruct i1 with a C grade (Runs well). Expected decode speed: 59.0 tok/s.
How much VRAM does EXAONE 3.5 7.8B Instruct i1 need?
EXAONE 3.5 7.8B Instruct i1 (7.800000190734863B parameters) requires approximately 8.9 GB of memory with Q4_K_M quantization.
What is the best quantization for EXAONE 3.5 7.8B Instruct i1?
The recommended quantization for EXAONE 3.5 7.8B Instruct i1 is Q4_K_M, which balances quality and memory efficiency.
What speed will EXAONE 3.5 7.8B Instruct i1 run at on RTX 4000 Ada 20GB?
On RTX 4000 Ada 20GB, EXAONE 3.5 7.8B Instruct i1 achieves approximately 59.0 tokens per second decode speed with a time-to-first-token of 3280ms using Q4_K_M quantization.
Can RTX 4000 Ada 20GB run EXAONE 3.5 7.8B Instruct i1 for coding?
For coding workloads, EXAONE 3.5 7.8B Instruct i1 on RTX 4000 Ada 20GB receives a C grade with 59.0 tok/s and 211K context.
What context window can EXAONE 3.5 7.8B Instruct i1 use on RTX 4000 Ada 20GB?
On RTX 4000 Ada 20GB, EXAONE 3.5 7.8B Instruct i1 can safely use up to 211K tokens of context. The model's official context limit is —, but available memory constrains the safe maximum.
Embed this result▼
Paste this snippet into any page to show a live fit card.
<iframe src="https://willitrunai.com/embed/hf-mradermacher--exaone-3-5-7-8b-instruct-i1-gguf-on-rtx-4000-ada-20gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>
Preview: