Local AI Advisor

Can Desktop PC · Intel Arc Pro B60 24GB · 32 GB RAM run Kolibri-1 (78B-A3.5B)?

24 GB VRAM + 32 GB RAM · 456 GB/s memory bandwidth · Kolibri-1 (78B-A3.5B) has 78.1B parameters (3.46B active per token).

Short answer

It loads, but it's borderline even for chat.

Best quantization for chat: Q4_K_M (45 GB of weights) on LM Studio. Generation: 19–30 tok/s, leaving about 0.0 GB free for your other apps.

How it rates for each workload

Same model and machine, different jobs. Agents and long documents send far bigger prompts than chat, so they need much more speed and memory.

WorkloadVerdictQuantGenerationFirst replyContext
Chatwith Open WebUIBorderlineQ4_K_M19–30 tok/s≈3 s8KDetails
Coding questionswith ContinueDoes Not Fit————Details
Coding in a repositorywith AiderDoes Not Fit————Details
Agentic codingwith OpenCodeDoes Not Fit————Details
Long documents (64K)with Open WebUIDoes Not Fit————Details

Speeds are bandwidth-based estimates; no direct benchmark exists for this pair. “First reply” is the time to read the workload’s opening prompt. How we calculate

Different tool, context size or workload?

These pages use typical settings. The full check models your exact setup.