codellama · text

CodeLlama-70b-Python-hf

codellama/CodeLlama-70b-Python-hf

CodeLlama-70b-Python-hf at Q4_K_M is exactly 41,423,092,192 bytes (38.58 GiB / 41.42 GB) — an effective 4.804 bits per weight, not the nominal 4. Its KV cache at 32K is 10.00 GiB.

From the file· summed from 1 file(s)From the file· KV per layer
Parameters
69.0B
Architecture
llama
80 layers
Context
4,096
native (config.json)
License
llama2

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
I1-IQ1_S13.54 GiB14,535,672,3841.686—mradermacher
I1-IQ1_M14.85 GiB15,943,385,6641.849—mradermacher
I1-IQ2_XXS17.03 GiB18,289,574,4642.121—mradermacher
I1-IQ2_XS18.94 GiB20,334,297,6642.358—mradermacher
I1-IQ2_S19.89 GiB21,354,460,7362.477—mradermacher
I1-IQ2_M21.64 GiB23,231,411,7762.694—mradermacher
Q2_K23.71 GiB25,462,587,8722.953—TheBloke
Q2_K23.71 GiB25,462,588,7362.953—mradermacher
I1-Q2_K23.71 GiB25,462,588,9922.953—mradermacher
I1-IQ3_XXS24.76 GiB26,581,612,0963.083—mradermacher
IQ3_XS26.37 GiB28,315,138,3683.284—mradermacher
I1-IQ3_XS26.37 GiB28,315,138,6243.284—mradermacher
Q3_K_S27.86 GiB29,919,458,7843.470—TheBloke
Q3_K_S27.86 GiB29,919,459,6483.470—mradermacher
IQ3_S27.86 GiB29,919,459,6483.470—mradermacher
I1-IQ3_S27.86 GiB29,919,459,9043.470—mradermacher
I1-Q3_K_S27.86 GiB29,919,459,9043.470—mradermacher
IQ3_M28.82 GiB30,944,442,6883.589—mradermacher
I1-IQ3_M28.82 GiB30,944,442,9443.589—mradermacher
Q3_K_M30.99 GiB33,274,901,9843.859—TheBloke
Q3_K_M30.99 GiB33,274,902,8483.859—mradermacher
I1-Q3_K_M30.99 GiB33,274,903,1043.859—mradermacher
Q3_K_L33.67 GiB36,148,000,2244.192—TheBloke
Q3_K_L33.67 GiB36,148,001,0884.192—mradermacher
I1-Q3_K_L33.67 GiB36,148,001,3444.192—mradermacher
I1-IQ4_XS34.30 GiB36,829,998,6564.272—mradermacher
IQ4_XS34.64 GiB37,197,000,0004.314—mradermacher
Q4_036.20 GiB38,872,431,0724.508—TheBloke
I1-Q4_036.34 GiB39,019,232,8324.526—mradermacher
Q4_K_S36.55 GiB39,249,918,4324.552—TheBloke
Q4_K_S36.55 GiB39,249,919,2964.552—mradermacher
I1-Q4_K_S36.55 GiB39,249,919,5524.552—mradermacher
Q4_K_M38.58 GiB41,423,092,1924.804—TheBloke
Q4_K_M38.58 GiB41,423,093,0564.804—mradermacher
I1-Q4_K_M38.58 GiB41,423,093,3124.804—mradermacher
Q5_044.20 GiB47,461,595,6165.505—TheBloke
Q5_K_S44.20 GiB47,461,595,6165.505—TheBloke
Q5_K_S44.20 GiB47,461,596,4805.505—mradermacher
I1-Q5_K_S44.20 GiB47,461,596,7365.505—mradermacher
Q5_K_M45.41 GiB48,753,965,5365.654—TheBloke

KV cache by context

computed per layer
ContextKV cache (f16)●Flat formulaOverstated byFull / windowed / recurrent
4,0961.25 GiB1.25 GiB—80 / 0 / 0
8,1922.50 GiB2.50 GiB—80 / 0 / 0
16,3845.00 GiB5.00 GiB—80 / 0 / 0
32,76810.00 GiB10.00 GiB—80 / 0 / 0
65,53620.00 GiB20.00 GiB—80 / 0 / 0
131,07240.00 GiB40.00 GiB—80 / 0 / 0

Compare with

same modality, comparable size
vs Qwen3-Coder-30B-A3B-Instructvs Qwen3.6-27Bvs Qwen3.8-27Bvs gemma-4-12B-it-qat-q4_0-unquantizedvs gemma-4-E4B-itvs Qwen3-30B-A3B-Thinking-2507All alternatives →Will it fit on my card? →

Will it run on your card?

full quant x context sweep
By VRAM:8GB12GB16GB24GB32GB48GB
Radeon RX 6500 XT 4GBGeForce RTX 3050 6GBGeForce RTX 5050 8GBGeForce RTX 3080 10GBGeForce RTX 2080 Ti 11GBGeForce RTX 5070 12GBGeForce RTX 5060 Ti 16GBApple M3 Pro 18GBGeForce RTX 3080 Ti 20GBGeForce RTX 5090 D V2 24GBGeForce RTX 5090 D 32GBApple M5 Max 36GBApple M5 Pro 48GBApple M5 Max 64GBApple M3 Ultra 96GBApple M5 Max 128GBApple M2 Ultra 192GBApple M3 Ultra 256GBApple M3 Ultra 512GB

Why other calculators give a different number

A parameters × bits ÷ 8 estimate puts Q4_K_M at roughly 36.13 GiB. The real file is 38.58 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

from config.json
Layers
80
Attention heads
64
KV heads
8
Head dim
128
Hidden size
8192
Vocab
32,016
Sliding window
none
SWA period
—
MLA
no
Experts
—
Experts per token
—
use_sliding_window
—

Questions people ask

How much VRAM does CodeLlama-70b-Python-hf need?
Q4_K_M is exactly 41,423,092,192 bytes (38.58 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
How large is CodeLlama-70b-Python-hf's KV cache?
10.00 GiB at 32K context with an f16 cache, computed per layer. Quantizing the cache to q8_0 roughly halves it, which is often the difference between a context length fitting and not.
Which quantization of CodeLlama-70b-Python-hf should I use?
Q4_K_M is the usual default. Pick the largest quantization that fits your card at the context you actually need — the table above gives exact sizes for every one published.