Kwaipilot_KAT-Coder-V2.5-Dev

Kwaipilot family · 35.0B params

Evidence: 0 sourced runs · 1 awaiting review · 1 independent source · first recorded Aug 15, 2026. What these labels mean.

What the Kwaipilot_KAT-Coder-V2.5-Dev numbers mean

The fastest result on our board is 42.2 output tok/s on an RTX 4070 at Q4_K_L with a 262144-token context. That is about 23.7 ms per generated token — 500 tokens takes roughly 11.9 seconds. Prompt processing on the same record runs at 64.4 tok/s, so a 2,000-token prompt is read in about 31.0 seconds. Peak VRAM was measured at 4.4 GB.

How much VRAM does Kwaipilot_KAT-Coder-V2.5-Dev need?

Kwaipilot_KAT-Coder-V2.5-Dev has 35.0B 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.

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.

Models with a similar footprint

Ranked by distance from Kwaipilot_KAT-Coder-V2.5-Dev's parameter count. Full benchmark board.

All benchmarks Methodology