Flint-40B-A2.5B · Q6_K
This page aggregates 1 real-world runs of Flint-40B-A2.5B (Q6_K) on NVIDIA H20 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
79.46 tok/s
Decode
Decode speed
372.23 tok/s
Prefill
Prefill speed
1.4 s
TTFT
Time to first token
85.68 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
14 days ago
Last verified
Performance
- vLLM · Q4_K_M Decode 94.37 · Prefill 330.77 ·
- vLLM · Q6_K (current)Decode 79.46 · Prefill 372.23 ·
- Ollama · Q3_K_S Decode 104.76 · Prefill 539.82 ·
Core figures
- Decode (avg)
- 79.46 tok/s
- Prefill (avg)
- 372.23 tok/s
- TTFT (avg)
- 1.4 s
- VRAM (avg)
- 85.68 GB
- MTP acceptance rate
- —
- TTFB
- 0.64 GB
- Power draw
- 324.1 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Flint-40B-A2.5B
- Quantization
- Q6_K
- Framework
- vLLM
- Version
- v0.32
- Context length
- 8192 tokens
- Batch size
- 8
- GPU layers
- 999
- Flash Attention
- Off
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA H20
- Nominal VRAM
- 97 GB
- Measured VRAM (avg)
- 85.68 GB
- OS
- Ubuntu 24.04
- Driver
- 570.86
- CUDA
- 12.6
- Power draw
- 324.1 W
Get started
Original model resources
Base model resources may not match this quantization. Use the verified deployment weights above when available.
Sources & evidence
1 measured records in total, each traceable to its original source
v0.32 · Ubuntu 24.04 · CUDA 12.6 · 8192 ctx
79.46 tok/s
Decode
372.23 tok/s
Prefill
1.4 s
TTFT
85.68 GB
VRAM
—
MTP
324.1 W
Power draw
L2 batch seed 2026-10-22: vLLM on hw_id=56 running model_id=92 at Q6_K