Reka AItransformer

Reka Flash

21B parameters • 12.8GB VRAM (Q4 estimate) • 128,000 context

Estimate · Q4_K_M · 4k context · batch 1. For dense 7B–70B this sits ~5–8% above the Q4_K_M GGUF file. Not peak runtime VRAM and not a measured bench.

Specifications

Parameters

21B

VRAM (Q4 est.)

12.8 GB

VRAM (FP16)

42 GB

Context Window

128,000

Architecture

transformer

License

Reka TOS

Finding cloud alternatives...

Run Reka Flash on Mac Mini M4

~12.8GB VRAM needed at Q4. Mac Mini M4 has 16GB — buy hardware or rent cloud.

Amazon eBay Compare cloud prices

We may earn a commission from cloud and hardware partners at no extra cost to you.

Find the cheapest GPU that can run Reka Flash

Compatibility Lab — check every GPU × quantization combination

→

Run Reka Flash on...

GB200 NVL72✅ Fits Lambda Labs 1-Click Cluster (8x H100)✅ Fits Google Cloud A3 (8x H100)✅ Fits Azure ND H100 v5 (8x H100)✅ Fits AWS EC2 P5 Instance (8x H100)✅ Fits NVIDIA DGX H100✅ Fits

VRAM Calculator→GPU Comparison→Cloud GPU Pricing→

← Back to Models

Reka AItransformer

Reka Flash

21B parameters • 12.8GB VRAM (Q4 estimate) • 128,000 context

Estimate · Q4_K_M · 4k context · batch 1. For dense 7B–70B this sits ~5–8% above the Q4_K_M GGUF file. Not peak runtime VRAM and not a measured bench.