thesby · vision language

Qwen2.5-VL-7B-NSFW-Caption-V3

thesby/Qwen2.5-VL-7B-NSFW-Caption-V3

Qwen2.5-VL-7B-NSFW-Caption-V3 at Q4_K_M is exactly 4,683,072,384 bytes (4.36 GiB / 4.68 GB) — an effective 4.919 bits per weight, not the nominal 4.

From the file· summed from 1 file(s)
Parameters
7.6B
Architecture
qwen2vl
Context
—
native (config.json)
License
apache-2.0

Shipped quantizations

● exact bytes, summed from published files
QuantSize●Exact bytes●Effective bpw●Tensors●Publisher
IQ2_M2.59 GiB2,780,341,1202.921—bartowski
IQ2_M2.59 GiB2,780,341,1202.921—Hi821119
Q2_K2.81 GiB3,015,938,9443.168—Hi821119
Q2_K2.81 GiB3,015,938,9443.168—bartowski
Q2_K2.81 GiB3,015,939,1363.168—mradermacher
IQ3_XXS2.90 GiB3,114,513,2803.272—bartowski
IQ3_XXS2.90 GiB3,114,513,2803.272—Hi821119
IQ3_XS3.12 GiB3,346,254,7203.515—Hi821119
IQ3_XS3.12 GiB3,346,254,7203.515—bartowski
Q3_K_S3.25 GiB3,492,367,2323.669—Hi821119
Q3_K_S3.25 GiB3,492,367,2323.669—bartowski
Q3_K_S3.25 GiB3,492,367,4243.669—mradermacher
Q2_K_L3.30 GiB3,548,162,9443.727—Hi821119
Q2_K_L3.30 GiB3,548,162,9443.727—bartowski
IQ3_M3.33 GiB3,574,010,7523.754—bartowski
IQ3_M3.33 GiB3,574,010,7523.754—Hi821119
Q3_K_M3.55 GiB3,808,390,0164.001—bartowski
Q3_K_M3.55 GiB3,808,390,0164.001—Hi821119
Q3_K_M3.55 GiB3,808,390,2084.001—mradermacher
Q3_K_L3.81 GiB4,088,458,1124.295—bartowski
Q3_K_L3.81 GiB4,088,458,1124.295—Hi821119
Q3_K_L3.81 GiB4,088,458,3044.295—mradermacher
IQ4_XS3.93 GiB4,218,471,2964.431—bartowski
IQ4_XS3.93 GiB4,218,471,2964.431—Hi821119
IQ4_XS3.96 GiB4,250,297,4084.465—mradermacher
IQ4_NL4.13 GiB4,437,812,0964.662—bartowski
IQ4_NL4.13 GiB4,437,812,0964.662—Hi821119
Q4_04.14 GiB4,444,119,9364.668—bartowski
Q4_04.14 GiB4,444,119,9364.668—Hi821119
Q4_K_S4.15 GiB4,457,767,8084.683—bartowski
Q4_K_S4.15 GiB4,457,767,8084.683—Hi821119
Q4_K_S4.15 GiB4,457,768,0004.683—mradermacher
Q4_K_M4.36 GiB4,683,072,3844.919—Hi821119
Q4_K_M4.36 GiB4,683,072,3844.919—bartowski
Q4_K_M4.36 GiB4,683,072,5764.919—mradermacher
Q4_14.54 GiB4,873,282,4325.119—bartowski
Q4_14.54 GiB4,873,282,4325.119—Hi821119
Q4_K_L4.74 GiB5,087,562,6245.344—Hi821119
Q4_K_L4.74 GiB5,087,562,6245.344—bartowski
Q5_K_S4.95 GiB5,315,175,2965.583—Hi821119

Compare with

same modality, comparable size
vs Qwen3.5-9Bvs gemma-4-12B-itvs Qwen3.5-4Bvs Qwythos-9B-Claude-Mythos-5-1Mvs Qwen3.5-0.8Bvs gemma-4-E2B-itAll 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 3.99 GiB. The real file is 4.36 GiB, because a quantization is a mixture and some tensors are always kept at higher precision.

Architecture

Architecture unavailable — this repository is gated and no ungated mirror was found. Exact file sizes above are still authoritative; only the KV math needs the config.

Questions people ask

How much VRAM does Qwen2.5-VL-7B-NSFW-Caption-V3 need?
Q4_K_M is exactly 4,683,072,384 bytes (4.36 GiB) in weights. Add the KV cache, which depends on your context length, plus roughly half a gigabyte of runtime overhead.
Which quantization of Qwen2.5-VL-7B-NSFW-Caption-V3 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.