Beam-1.0

Beam family · 501.0B params

4recorded runs

58.9best tok/s

4verified

Evidence: 4 sourced runs · 1 independent source · first recorded Oct 30, 2026. What these labels mean.

What the Beam-1.0 numbers mean

The fastest verified result on our board is 58.9 output tok/s on an DGX Spark x2 at EXL3 3 bpw with a 46000-token context. That is about 17.0 ms per generated token — 500 tokens takes roughly 8.5 seconds.

How much VRAM does Beam-1.0 need?

Beam-1.0 has 501.0B parameters. These estimates use the same formula as the VRAM calculator: weights, plus about 15% for a moderate-context KV cache and runtime overhead, plus a small engine floor.

Quantisation	Estimated VRAM	Smallest card that fits
Q4_K_M (4-bit)	324.9 GB	multi-GPU or CPU offload
Q8_0 (8-bit)	613.0 GB	multi-GPU or CPU offload
FP16 (16-bit)	1153.1 GB	multi-GPU or CPU offload

Long contexts and large batches raise the real number — read the smallest card as "possible with a reduced context". Measured throughput is on the benchmark board.

Fastest GPUs for this model (best tok/s)

DGX Spark x258.9

Models with a similar footprint

Model	Params	Q4_K_M weights
DeepSeek-V4.1-Flash	552B	310.5 GB
DeepSeek-V4.1-Flash-1017	552B	310.5 GB
DeepSeek-V4.1-Flash-1017-Sushi-STQ	552B	310.5 GB
MiniMax M3.1 Flash	428B	240.8 GB
GLM-5.3-Flash	321B	180.6 GB
DeepSeek-V4-Flash-Vision-Exp	305B	171.6 GB

Ranked by distance from Beam-1.0's parameter count. Full benchmark board.

All benchmarks Methodology

GPU	Model	Quant	Backend	Ctx	tok/s	prompt tok/s	VRAM GB	Source
DGX Spark x2 · 128 GB	Beam-1.0	EXL3 3 bpw	tensorfold recipe v0.2.0	46000	58.9 ✓	—	—	src
DGX Spark x2 · 128 GB	Beam-1.0	EXL3 3 bpw	tensorfold recipe v0.2.0	2048	45.9 ✓	—	—	src
DGX Spark x2 · 128 GB	Beam-1.0	EXL3 3 bpw	tensorfold recipe v0.2.0	32768	41.6 ✓	—	—	src
DGX Spark x2 · 128 GB	Beam-1.0	EXL3 3 bpw	tensorfold recipe v0.2.0	16384	31.5 ✓	—	—	src