RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can I run Unlimited OCR 2 locally?

Unlimited OCR 2 by baidu has 3.3B parameters (1.1B active per token, 64 experts, 6 used per token). At 4-bit with an 8K context it needs about 3.0 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track. See all models that fit in 8 GB of VRAM.

3.0 GBQ4_K_M · 8K context
RTX 4060smallest consumer GPU that fits
M1 16GBsmallest Mac that fits
128Kmax context
MHAattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from AtomicChat/Unlimited-OCR-2-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K)rough2.0 GB2.3 GB3.7 GB9.3 GB
3-bit (Q3_K_M)usable2.4 GB2.6 GB4.0 GB9.7 GB
4-bit (Q4_K_M)good, most popular2.8 GB3.0 GB4.4 GB10.1 GB
5-bit (Q5_K_M)very good3.1 GB3.3 GB4.7 GB10.4 GB
6-bit (Q6_K) •near lossless3.4 GB3.7 GB5.1 GB10.7 GB
8-bit (Q8_0) •effectively lossless3.8 GB4.0 GB5.4 GB11.1 GB
16-bit (F16/BF16) •original weights6.4 GB6.6 GB8.0 GB13.6 GB

Which GPUs and Macs run Unlimited OCR 2?

Best quality quant that fits at 8K context. "Tight" = within 10% of usable memory. Speed is a ceiling from memory bandwidth (real speed is typically 50–80% of it).

HardwareMemoryFits?Best quantNeedsMax tok/s (est.)
RTX 3060 12GB12 GBFitsF166.6 GB178details
RTX 4060 Ti 16GB16 GBFitsF166.6 GB142details
RTX 309024 GBFitsF166.6 GB463details
RTX 409024 GBFitsF166.6 GB499details
RTX 509032 GBFitsF166.6 GB886details
RTX 5070 Ti16 GBFitsF166.6 GB443details
RX 7900 XTX24 GBFitsF166.6 GB475details
Mac M4 Pro 48GB48 GBFitsF166.6 GB135details
Mac M4 Max 128GB128 GBFitsF166.6 GB270details
Mac M3 Ultra 512GB512 GBFitsF166.6 GB405details
DGX Spark128 GBFitsF166.6 GB135details
RTX 407012 GBFitsF166.6 GB249details
RTX 5060 Ti 16GB16 GBFitsF166.6 GB222details
RTX 40608 GBFitsF166.6 GB135details
Arc B58012 GBFitsF166.6 GB226details
RTX 3060 Ti8 GBFitsF166.6 GB222details
RTX 30708 GBFitsF166.6 GB222details
RTX 3080 10GB10 GBFitsF166.6 GB376details
RTX 3080 12GB12 GBFitsF166.6 GB451details
RTX 3090 Ti24 GBFitsF166.6 GB499details
RTX 4060 Ti 8GB8 GBFitsF166.6 GB142details
RTX 4070 Super12 GBFitsF166.6 GB249details
RTX 4070 Ti12 GBFitsF166.6 GB249details
RTX 4070 Ti Super16 GBFitsF166.6 GB332details
RTX 408016 GBFitsF166.6 GB355details
RTX 4080 Super16 GBFitsF166.6 GB364details
RTX 50608 GBFitsF166.6 GB222details
RTX 507012 GBFitsF166.6 GB332details
RTX 508016 GBFitsF166.6 GB475details
2x RTX 30902×24 GBFitsF166.6 GB463details
2x RTX 40902×24 GBFitsF166.6 GB499details
2x RTX 50902×32 GBFitsF166.6 GB886details
RTX A600048 GBFitsF166.6 GB380details
RTX 6000 Ada48 GBFitsF166.6 GB475details
RTX PRO 6000 Blackwell96 GBFitsF166.6 GB886details
L40S48 GBFitsF166.6 GB427details
A100 40GB40 GBFitsF166.6 GB769details
A100 80GB80 GBFitsF166.6 GB1009details
H100 PCIe80 GBFitsF166.6 GB989details
H100 SXM80 GBFitsF166.6 GB1657details
H200141 GBFitsF166.6 GB2374details
RX 76008 GBFitsF166.6 GB142details
RX 7700 XT12 GBFitsF166.6 GB214details
RX 7800 XT16 GBFitsF166.6 GB309details
RX 7900 GRE16 GBFitsF166.6 GB285details
RX 7900 XT20 GBFitsF166.6 GB396details
RX 907016 GBFitsF166.6 GB317details
RX 9070 XT16 GBFitsF166.6 GB317details
Ryzen AI Max+ 395 128GB128 GBFitsF166.6 GB127details
Arc A770 16GB16 GBFitsF166.6 GB277details
Mac M1 16GB16 GBFitsF166.6 GB34details
Mac M2 24GB24 GBFitsF166.6 GB49details
Mac M3 24GB24 GBFitsF166.6 GB49details
Mac M4 16GB16 GBFitsF166.6 GB59details
Mac M4 32GB32 GBFitsF166.6 GB59details
Mac M1 Pro 32GB32 GBFitsF166.6 GB99details
Mac M3 Pro 36GB36 GBFitsF166.6 GB74details
Mac M4 Pro 24GB24 GBFitsF166.6 GB135details
Mac M4 Pro 64GB64 GBFitsF166.6 GB135details
Mac M1 Max 64GB64 GBFitsF166.6 GB198details
Mac M2 Max 96GB96 GBFitsF166.6 GB198details
Mac M3 Max 128GB128 GBFitsF166.6 GB198details
Mac M4 Max 64GB64 GBFitsF166.6 GB270details
Mac M2 Ultra 192GB192 GBFitsF166.6 GB396details
Mac M3 Ultra 256GB256 GBFitsF166.6 GB405details

Questions

How much VRAM does Unlimited OCR 2 need?

About 3.0 GB at 4-bit (Q4_K_M) with an 8K context: 1.9 GB of weights (estimated), 0.5 GB of KV cache and 0.6 GB of runtime overhead. At 8-bit it needs about 4.0 GB.

What is the smallest GPU that runs Unlimited OCR 2?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RTX 4060 (8 GB).

Can I run Unlimited OCR 2 on a Mac?

Yes. At Q4_K_M with 8K context it fits on a Mac M1 16GB or larger, using the default macOS GPU memory limit.

Similar models

Same family or a similar parameter count, with the VRAM each needs at Q4_K_M and 8K context.

Unlimited OCR

3.3B · 2.9 GB at 4-bit

Qianfan OCR

4.7B · 4.4 GB at 4-bit

ERNIE 4.5 21B A3B Thinking

22B · 13.5 GB at 4-bit

DeepSeek OCR

3.3B · 2.9 GB at 4-bit

deepseek vl2 tiny

3.4B · 3.0 GB at 4-bit

DeepSeek OCR 2

3.4B · 2.9 GB at 4-bit

Model facts

Hugging Face repobaidu/Unlimited-OCR-2
Parameters3,343,978,497 (3.3B) · ~1.1B active (computed from config)
Layers / hidden size12 / 1280
AttentionMHA; 10 heads, 10 KV heads, head dim 128
KV cache per token (FP16)60.0 KiB
Max context131,072 tokens
Licensemit
Released on HFOct 26, 2026
Downloads (30d) / likes559,340 / 704

Sources: Hugging Face model API · config.json (baidu/Unlimited-OCR-2) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.