Architecture truth card — Kimi-K2.7-Code
← Open the interactive VRAM & cost calculator on the homepage
Exact VRAM matrix — 3 precisions × 3 context lengths
| Precision | 8192 tokens ctx | 32768 tokens ctx | 131072 tokens ctx |
|---|---|---|---|
| FP16/BF16 16-bit | 1912.71 / 0.54 / 97.14 Total 2010.38 GiB | 1912.71 / 2.14 / 97.14 Total 2011.99 GiB | 1912.71 / 8.58 / 97.14 Total 2018.43 GiB |
| FP8/INT8 8-bit | 956.36 / 0.54 / 49.32 Total 1006.21 GiB | 956.36 / 2.14 / 49.32 Total 1007.82 GiB | 956.36 / 8.58 / 49.32 Total 1014.25 GiB |
| INT4/AWQ 4.5-bit | 537.95 / 0.54 / 28.40 Total 566.88 GiB | 537.95 / 2.14 / 28.40 Total 568.49 GiB | 537.95 / 8.58 / 28.40 Total 574.93 GiB |
Each cell: Weights GiB / GQA KV Cache GiB / Runtime Overhead GiB; bold line = Total VRAM GiB. Concurrency = 1, FP16 KV cache (2 bytes), overhead = 1.5 + 0.05 × weights GiB.
Terminal speed-math — no install needed
Re-run this exact sizing from your own terminal:
curl -fsSL https://gpuvram.pages.dev/vram-cli.py | python3 - kimi-k2-7-code --ctx 32768
Live cheapest GPU configurations
| Scenario | Cheapest fitting rig (top 3) | Total VRAM | Lowest $/hr | 730 h/mo | Deploy |
|---|---|---|---|---|---|
| INT4 @ 32K ctx | 4x H200 NVL 143GB | 572 GiB | $2.00 | $1460.00 | Deploy on RunPodCheck Vast.ai Verified |
| INT4 @ 32K ctx | 4x MI300X 192GB | 768 GiB | $2.00 | $1460.00 | Deploy on RunPodCheck Vast.ai Verified |
| INT4 @ 32K ctx | 8x H200 NVL 143GB | 1144 GiB | $4.00 | $2920.00 | Deploy on RunPodCheck Vast.ai Verified |
| FP8 @ 32K ctx | 8x H200 NVL 143GB | 1144 GiB | $4.00 | $2920.00 | Deploy on RunPodCheck Vast.ai Verified |
| FP8 @ 32K ctx | 8x MI300X 192GB | 1536 GiB | $4.00 | $2920.00 | Deploy on RunPodCheck Vast.ai Verified |
| FP8 @ 32K ctx | 4x B300 288GB | 1152 GiB | $27.76 | $20264.80 | Deploy on RunPodCheck Vast.ai Verified |
| FP16 @ 32K ctx | 8x B300 288GB | 2304 GiB | $55.52 | $40529.60 | Deploy on RunPodCheck Vast.ai Verified |
What over-provisioning costs you
Running Kimi-K2.7-Code production traffic at INT4 / 32K context on the cheapest fitting rig (4x H200 NVL 143GB) bills $1460.00 per month (730 h). Defaulting to full FP16 forces a larger rig — 8x B300 288GB at $40529.60/mo. Paying $40529.60 instead of $1460.00 every single month is the price of skipping 4-bit AWQ quantization; if your quality budget allows INT4, that gap is pure waste.