RunsOnMy ModelsGPUs & MacsAPI pricesVRAM calculatorNew

Can I run GLiDE 27B locally?

GLiDE 27B by fastino has 26B parameters. At 4-bit with an 8K context it needs about 16.2 GB of GPU memory. It fits on 58 of 65 GPUs and Macs we track.

16.2 GBQ4_K_M · 8K context
RX 7900 XTsmallest consumer GPU that fits
M4 32GBsmallest Mac that fits
256Kmax context
GQAattention type

Memory needed by quant and context length

GB of GPU memory = weights + KV cache (FP16) + runtime overhead. Weights use parameter count × bits per weight. How we calculate.

QuantQuality4K ctx8K ctx32K ctx128K ctx
2-bit (Q2_K)rough10.2 GB10.5 GB12.0 GB18.0 GB
3-bit (Q3_K_M)usable13.0 GB13.2 GB14.7 GB20.7 GB
4-bit (Q4_K_M)good, most popular16.0 GB16.2 GB17.7 GB23.7 GB
5-bit (Q5_K_M)very good18.4 GB18.7 GB20.2 GB26.2 GB
6-bit (Q6_K)near lossless21.0 GB21.3 GB22.8 GB28.8 GB
8-bit (Q8_0)effectively lossless26.9 GB27.2 GB28.7 GB34.7 GB
16-bit (F16/BF16)original weights49.7 GB50.0 GB51.5 GB57.5 GB

Which GPUs and Macs run GLiDE 27B?

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 GBTightQ2_K10.5 GB37details
RTX 4060 Ti 16GB16 GBFitsQ3_K_M13.2 GB23details
RTX 309024 GBFitsQ5_K_M18.7 GB50details
RTX 409024 GBFitsQ5_K_M18.7 GB54details
RTX 509032 GBFitsQ8_027.2 GB65details
RTX 5070 Ti16 GBFitsQ3_K_M13.2 GB70details
RX 7900 XTX24 GBFitsQ5_K_M18.7 GB52details
Mac M4 Pro 48GB48 GBFitsQ8_027.2 GB9.9details
Mac M4 Max 128GB128 GBFitsF1650.0 GB10details
Mac M3 Ultra 512GB512 GBFitsF1650.0 GB16details
DGX Spark128 GBFitsF1650.0 GB5.2details
RTX 407012 GBTightQ2_K10.5 GB52details
RTX 5060 Ti 16GB16 GBFitsQ3_K_M13.2 GB35details
RTX 40608 GBNo———details
Arc B58012 GBTightQ2_K10.5 GB47details
RTX 3060 Ti8 GBNo———details
RTX 30708 GBNo———details
RTX 3080 10GB10 GBNo———details
RTX 3080 12GB12 GBTightQ2_K10.5 GB93details
RTX 3090 Ti24 GBFitsQ5_K_M18.7 GB54details
RTX 4060 Ti 8GB8 GBNo———details
RTX 4070 Super12 GBTightQ2_K10.5 GB52details
RTX 4070 Ti12 GBTightQ2_K10.5 GB52details
RTX 4070 Ti Super16 GBFitsQ3_K_M13.2 GB53details
RTX 408016 GBFitsQ3_K_M13.2 GB56details
RTX 4080 Super16 GBFitsQ3_K_M13.2 GB58details
RTX 50608 GBNo———details
RTX 507012 GBTightQ2_K10.5 GB69details
RTX 508016 GBFitsQ3_K_M13.2 GB75details
2x RTX 30902×24 GBFitsQ8_027.2 GB34details
2x RTX 40902×24 GBFitsQ8_027.2 GB36details
2x RTX 50902×32 GBFitsF1650.0 GB34details
RTX A600048 GBFitsQ8_027.2 GB28details
RTX 6000 Ada48 GBFitsQ8_027.2 GB35details
RTX PRO 6000 Blackwell96 GBFitsF1650.0 GB34details
L40S48 GBFitsQ8_027.2 GB31details
A100 40GB40 GBFitsQ8_027.2 GB56details
A100 80GB80 GBFitsF1650.0 GB39details
H100 PCIe80 GBFitsF1650.0 GB38details
H100 SXM80 GBFitsF1650.0 GB64details
H200141 GBFitsF1650.0 GB92details
RX 76008 GBNo———details
RX 7700 XT12 GBTightQ2_K10.5 GB44details
RX 7800 XT16 GBFitsQ3_K_M13.2 GB49details
RX 7900 GRE16 GBFitsQ3_K_M13.2 GB45details
RX 7900 XT20 GBFitsQ4_K_M16.2 GB50details
RX 907016 GBFitsQ3_K_M13.2 GB50details
RX 9070 XT16 GBFitsQ3_K_M13.2 GB50details
Ryzen AI Max+ 395 128GB128 GBFitsF1650.0 GB4.9details
Arc A770 16GB16 GBFitsQ3_K_M13.2 GB44details
Mac M1 16GB16 GBTightQ2_K10.5 GB7.0details
Mac M2 24GB24 GBFitsQ3_K_M13.2 GB7.8details
Mac M3 24GB24 GBFitsQ3_K_M13.2 GB7.8details
Mac M4 16GB16 GBTightQ2_K10.5 GB12details
Mac M4 32GB32 GBFitsQ5_K_M18.7 GB6.5details
Mac M1 Pro 32GB32 GBFitsQ5_K_M18.7 GB11details
Mac M3 Pro 36GB36 GBFitsQ6_K21.3 GB7.0details
Mac M4 Pro 24GB24 GBFitsQ3_K_M13.2 GB21details
Mac M4 Pro 64GB64 GBFitsQ8_027.2 GB9.9details
Mac M1 Max 64GB64 GBFitsQ8_027.2 GB14details
Mac M2 Max 96GB96 GBFitsF1650.0 GB7.7details
Mac M3 Max 128GB128 GBFitsF1650.0 GB7.7details
Mac M4 Max 64GB64 GBFitsQ8_027.2 GB20details
Mac M2 Ultra 192GB192 GBFitsF1650.0 GB15details
Mac M3 Ultra 256GB256 GBFitsF1650.0 GB16details

Questions

How much VRAM does GLiDE 27B need?

About 16.2 GB at 4-bit (Q4_K_M) with an 8K context: 14.8 GB of weights (estimated), 0.5 GB of KV cache and 0.9 GB of runtime overhead. At 8-bit it needs about 27.2 GB.

What is the smallest GPU that runs GLiDE 27B?

At Q4_K_M and 8K context, the smallest single consumer GPU in our table that fits it is the RX 7900 XT (20 GB).

Can I run GLiDE 27B on a Mac?

Yes. At Q4_K_M with 8K context it fits on a Mac M4 32GB or larger, using the default macOS GPU memory limit.

Model facts

Hugging Face repofastino/GLiDE-27B
Parameters26,085,330,160 (26B)
Layers / hidden size64 / 5120
AttentionGQA (hybrid: 48 linear/state layers); 24 heads, 4 KV heads, head dim 256
KV cache per token (FP16)64.0 KiB
Max context262,144 tokens
Licenseother
Released on HFOct 19, 2026
Downloads (30d) / likes3,251 / 98

Sources: Hugging Face model API · config.json (fastino/GLiDE-27B) · GPU specs from vendor pages (list). Data fetched Nov 17, 2026. Spot an error? Report it.