Local AI Advisor

Can Desktop PC · RTX PRO 6000 Blackwell · 128 GB RAM run Recursive Flint (40B-A2.5B)?

96 GB VRAM + 128 GB RAM · 1792 GB/s memory bandwidth · Recursive Flint (40B-A2.5B) has 39.8B parameters (2.5B active per token).

Short answer

Yes, excellent for chat and comfortable or better for every workload we test.

Best quantization for chat: MXFP4 (20 GB of weights) on Ollama. Generation: 480–650 tok/s, leaving about 74 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 WebUIExcellentMXFP4480–650 tok/sunder 1 s8KDetails
Coding questionswith ContinueExcellentMXFP4450–610 tok/sunder 1 s16KDetails
Coding in a repositorywith AiderComfortableMXFP4370–500 tok/s≈1 s32KDetails
Agentic codingwith OpenCodeComfortableMXFP4350–470 tok/s≈1 s32KDetails
Long documents (64K)with Open WebUIExcellentMXFP4240–320 tok/s≈3 s64KDetails

Speeds are estimates calibrated against measured benchmarks on similar hardware. “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.