RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

Can I run Step3 VL 10B locally?

Step3 VL 10B by stepfun-ai has 10B parameters. At 4-bit with an 8K context it needs about 6.5 GB of GPU memory. It fits on 65 of 65 GPUs and Macs we track.

6.5 GBQ4_K_M · 8K context
RTX 4060smallest consumer GPU that fits
M1 16GBsmallest Mac that fits
64Kmax context
GQAattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use real file sizes from seanbailey518/Step3-VL-10B-GGUF where available (marked •), otherwise parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx
2-bit (Q2_K)rough4.8 GB5.4 GB8.7 GB
3-bit (Q3_K_M) •usable5.1 GB5.7 GB9.0 GB
4-bit (Q4_K_M) •good, most popular5.9 GB6.5 GB9.9 GB
5-bit (Q5_K_M) •very good6.7 GB7.3 GB10.6 GB
6-bit (Q6_K) •near lossless7.5 GB8.1 GB11.5 GB
8-bit (Q8_0) •effectively lossless9.4 GB9.9 GB13.3 GB
16-bit (F16/BF16) •original weights31.8 GB32.3 GB35.7 GB

Which GPUs and Macs run Step3 VL 10B?

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 GBFitsQ8_09.9 GB41details
RTX 4060 Ti 16GB16 GBFitsQ8_09.9 GB33details
RTX 309024 GBFitsQ8_09.9 GB107details
RTX 409024 GBFitsQ8_09.9 GB116details
RTX 509032 GBFitsQ8_09.9 GB206details
RTX 5070 Ti16 GBFitsQ8_09.9 GB103details
RX 7900 XTX24 GBFitsQ8_09.9 GB110details
Mac M4 Pro 48GB48 GBFitsQ8_09.9 GB31details
Mac M4 Max 128GB128 GBFitsF1632.3 GB17details
Mac M3 Ultra 512GB512 GBFitsF1632.3 GB25details
DGX Spark128 GBFitsF1632.3 GB8.3details
RTX 407012 GBFitsQ8_09.9 GB58details
RTX 5060 Ti 16GB16 GBFitsQ8_09.9 GB51details
RTX 40608 GBFitsQ4_K_M6.5 GB54details
Arc B58012 GBFitsQ8_09.9 GB52details
RTX 3060 Ti8 GBFitsQ4_K_M6.5 GB89details
RTX 30708 GBFitsQ4_K_M6.5 GB89details
RTX 3080 10GB10 GBFitsQ6_K8.1 GB113details
RTX 3080 12GB12 GBFitsQ8_09.9 GB105details
RTX 3090 Ti24 GBFitsQ8_09.9 GB116details
RTX 4060 Ti 8GB8 GBFitsQ4_K_M6.5 GB57details
RTX 4070 Super12 GBFitsQ8_09.9 GB58details
RTX 4070 Ti12 GBFitsQ8_09.9 GB58details
RTX 4070 Ti Super16 GBFitsQ8_09.9 GB77details
RTX 408016 GBFitsQ8_09.9 GB82details
RTX 4080 Super16 GBFitsQ8_09.9 GB85details
RTX 50608 GBFitsQ4_K_M6.5 GB89details
RTX 507012 GBFitsQ8_09.9 GB77details
RTX 508016 GBFitsQ8_09.9 GB110details
2x RTX 30902×24 GBFitsF1632.3 GB29details
2x RTX 40902×24 GBFitsF1632.3 GB31details
2x RTX 50902×32 GBFitsF1632.3 GB55details
RTX A600048 GBFitsF1632.3 GB23details
RTX 6000 Ada48 GBFitsF1632.3 GB29details
RTX PRO 6000 Blackwell96 GBFitsF1632.3 GB55details
L40S48 GBFitsF1632.3 GB26details
A100 40GB40 GBFitsF1632.3 GB47details
A100 80GB80 GBFitsF1632.3 GB62details
H100 PCIe80 GBFitsF1632.3 GB61details
H100 SXM80 GBFitsF1632.3 GB102details
H200141 GBFitsF1632.3 GB146details
RX 76008 GBFitsQ4_K_M6.5 GB57details
RX 7700 XT12 GBFitsQ8_09.9 GB50details
RX 7800 XT16 GBFitsQ8_09.9 GB72details
RX 7900 GRE16 GBFitsQ8_09.9 GB66details
RX 7900 XT20 GBFitsQ8_09.9 GB92details
RX 907016 GBFitsQ8_09.9 GB73details
RX 9070 XT16 GBFitsQ8_09.9 GB73details
Ryzen AI Max+ 395 128GB128 GBFitsF1632.3 GB7.8details
Arc A770 16GB16 GBFitsQ8_09.9 GB64details
Mac M1 16GB16 GBFitsQ6_K8.1 GB10details
Mac M2 24GB24 GBFitsQ8_09.9 GB11details
Mac M3 24GB24 GBFitsQ8_09.9 GB11details
Mac M4 16GB16 GBFitsQ6_K8.1 GB18details
Mac M4 32GB32 GBFitsQ8_09.9 GB14details
Mac M1 Pro 32GB32 GBFitsQ8_09.9 GB23details
Mac M3 Pro 36GB36 GBFitsQ8_09.9 GB17details
Mac M4 Pro 24GB24 GBFitsQ8_09.9 GB31details
Mac M4 Pro 64GB64 GBFitsF1632.3 GB8.3details
Mac M1 Max 64GB64 GBFitsF1632.3 GB12details
Mac M2 Max 96GB96 GBFitsF1632.3 GB12details
Mac M3 Max 128GB128 GBFitsF1632.3 GB12details
Mac M4 Max 64GB64 GBFitsF1632.3 GB17details
Mac M2 Ultra 192GB192 GBFitsF1632.3 GB24details
Mac M3 Ultra 256GB256 GBFitsF1632.3 GB25details

Questions

How much VRAM does Step3 VL 10B need?

About 6.5 GB at 4-bit (Q4_K_M) with an 8K context: 4.7 GB of weights (real GGUF file size), 1.1 GB of KV cache and 0.7 GB of runtime overhead. At 8-bit it needs about 9.9 GB.

What is the smallest GPU that runs Step3 VL 10B?

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 Step3 VL 10B 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.

Model facts

Hugging Face repostepfun-ai/Step3-VL-10B
Parameters10,171,750,144 (10B)
Layers / hidden size36 / 4096
AttentionGQA; 32 heads, 8 KV heads, head dim 128
KV cache per token (FP16)144.0 KiB
Max context65,536 tokens
Licenseapache-2.0
Released on HFJan 13, 2026
Downloads (30d) / likes4,972 / 418

Sources: Hugging Face model API · config.json (stepfun-ai/Step3-VL-10B) · GGUF files · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.