Can MacBook Pro M2 Max 96 GB run Recursive Flint (40B-A2.5B)?

96 GB unified · 400 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 LM Studio. Generation: 130–170 tok/s, leaving about 65 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.

Workload	Verdict	Quant	Generation	First reply	Context
Chatwith Open WebUI	Excellent	MXFP4	130–170 tok/s	under 1 s	8K	Details
Coding questionswith Continue	Excellent	MXFP4	120–160 tok/s	≈3 s	16K	Details
Coding in a repositorywith Aider	Comfortable	MXFP4	100–130 tok/s	≈9 s	32K	Details
Agentic codingwith OpenCode	Comfortable	MXFP4	90–120 tok/s	≈10 s	32K	Details
Long documents (64K)with Open WebUI	Excellent	MXFP4	65–85 tok/s	≈35 s	64K	Details

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

Other models for MacBook Pro M2 Max 96 GB

The best-rated models for chat on this machine.

Everything MacBook Pro M2 Max 96 GB can run

Cheapest machines for Recursive Flint (40B-A2.5B)

Hardware rated comfortable or better for chat, lowest price first.

Same chip, other memory: 32 GB, 64 GB

Different tool, context size or workload?

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

Open sourceCheck my exact setup