Pythia 70M Deduped — Hardware Requirements & GPU Compatibility
Chat
Pythia 70M Deduped is a 96M-parameter open language model from EleutherAI. It supports a context window of up to 2,048 tokens. At Q4_K_M it needs about 0.06 GB of VRAM — see which GPUs and Macs can run it below.
Runs great on a 24 GB GPU
Runs from 8 GB (Q4_K_M · 2K)Best on 24 GB (Q4_K_M · ~10920 tok/s · 2K)Macs from 8 GB (Q4_K_M · ~1195 tok/s)
Apache 2.0
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Pythia 70M Deduped
One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.
Recommended
ollama run hf.co/mradermacher/pythia-70m-deduped-i1-GGUF:Q4_K_MQ4_K_M fits fully on the RTX 4090 · 23.9 GB free · 2K context
Download: pythia-70m-deduped.i1-Q4_K_M.gguf0.05 GBmradermacher
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on RTX 3080 Ti (12 GB) (11 GB headroom) · Q4_K_M
- Generation speed
- ~9884 tok/s
- generation speed
- Cost per 1M output tokens
- $0.0034
- per 1M output tokens
How Much VRAM Does Pythia 70M Deduped Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 0.0 GB | — | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 2K context · ~16380 tok/s | |
| 0.1 GB | — | 4-bit · legacy — 4-bit legacy quantization | Fits: 2K context · ~13104 tok/s | |
| 0.1 GB | — | 3-bit small — 3-bit small quantization | Fits: 2K context · ~13104 tok/s | |
| 0.1 GB | — | 3-bit medium — 3-bit medium quantization | Fits: 2K context · ~13104 tok/s | |
| 0.1 GB | — | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 2K context · ~10920 tok/s | |
| 0.1 GB | — | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Fits: 2K context · ~9360 tok/s | |
| 0.1 GB | — | 6-bit, very good — 6-bit quantization, very good quality | Fits: 2K context · ~7280 tok/s | |
| 0.1 GB | — | 8-bit, near-lossless — 8-bit quantization, near-lossless | Fits: 2K context · ~5956 tok/s |
Selected Q4_K_M
4-bit medium quantization0.1GBat 2K context
- Weights
- 0.1 GB
- KV cache
- 0.0 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 23.9 GB free · ~10920 tok/s
≥50 tok/s needs ≥5 GB/s · RTX 5090 ~19413, RTX 3090 Ti ~10920
Which Hardware Can Run Pythia 70M Deduped?
Q4_K_M · 0.1 GBPythia 70M Deduped (Q4_K_M) requires 0.1 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 1+ GB is recommended. 52 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 163 devices with unified memory can run Pythia 70M Deduped, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB).
≤ 11 GB
≤ 11 GB tier →fits · 7.9 GB free · up to 2K ctx
16 GPUs · 1 Mac class · 2 device families
RTX 3080~8237Simulate output on RTX 3080RTX 2080 Ti~6673Simulate output on RTX 2080 TiRTX 3070 Ti~6590Simulate output on RTX 3070 TiM3*8 GB~1195Simulate output on iMac 24" M3 (8 GB)Orin Nano~1105Simulate output on NVIDIA Jetson Orin Nano 8GB (Super)iPhone~796Simulate output on iPhone 1720 GB
fits · 19.9 GB free · up to 2K ctx
1 GPU
RX 7900 XT~8000Simulate output on RX 7900 XT32–40 GB
fits · 31.9 GB free · up to 2K ctx
1 GPU · 3 Mac classes · 4 device families
RTX 5090~19413Simulate output on RTX 5090M5 Max36 GB~5367Simulate output on Mac Studio M5 Max (36 GB)M532 GB~1792Simulate output on MacBook Air 13" M5 (32 GB)Orin~2219Simulate output on NVIDIA Jetson AGX Orin 32GBSnapdragon X Elite (X1E-80-100)~1463Simulate output on Snapdragon X Elite Copilot+ PC
By VRAM tier · typical GPUs and Macs at Q4_K_M
Specifications
- Publisher
- EleutherAI
- Parameters
- 96M
- Architecture
- GPTNeoXForCausalLM
- Context Length
- 2,048 tokens
- Vocabulary Size
- 50,304
- Release Date
- 2023-02-13
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does Pythia 70M Deduped need?
Pythia 70M Deduped requires 0.1 GB of VRAM at Q4_K_M, or 0.2 GB at FP16.
VRAM = Weights + KV Cache + Overhead
Weights = 96M × 4.8 bits ÷ 8 = 0.1 GB
Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.
VRAM usage by quantization
What's the best quantization for Pythia 70M Deduped?
For Pythia 70M Deduped, Q4_K_M (0.1 GB) offers the best balance of quality and VRAM usage. Q5_K_M (0.1 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XS at 0.0 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Pythia 70M Deduped on a Mac?
Yes — MacBook Air 13" M3 (8 GB) and 145 other Macs can run Pythia 70M Deduped. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 0.1 GB of usable unified memory (RAM minus macOS overhead).
Can I run Pythia 70M Deduped locally?
Yes — Pythia 70M Deduped can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Pythia 70M Deduped?
At Q4_K_M, Pythia 70M Deduped can reach ~80000 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~10920 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 0.06 × 0.65 = ~86667 tok/s
Estimated speed at Q4_K_M (0.1 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Pythia 70M Deduped?
At Q4_K_M, the download is about 0.06 GB. The full-precision FP16 version is 0.19 GB. The smallest option (IQ2_XS) is 0.03 GB.
Which GPUs can run Pythia 70M Deduped?
52 consumer GPUs can run Pythia 70M Deduped at Q4_K_M. Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 52 GPUs have plenty of headroom for comfortable inference.
Which devices can run Pythia 70M Deduped?
163 devices with unified memory can run Pythia 70M Deduped at Q4_K_M (0.1 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.