Qwen3.6 27B Achieves 81 TPS Local Inference on M5 Max
max_paperclips · x · 2026-07-03
Shared test data reveals that running the Qwen 3.6 27B model on an Apple M5 Max chip can achieve an inference speed of 81 tokens/second, maintaining 63 TPS even when generating 11k tokens. This test was conducted using the MTPLX inference optimization tool, whose upcoming V2 release will further enhance local LLM inference performance.
More from Infra
- Anthropic's S-1 Reveals Over $600B in Infrastructure Commitments Across Google, Amazon, Microsoft — cedric_chee · 2026-10-26
- Anthropic S-1 reveals 48 cents of compute cost per dollar of Q2 revenue — nmasc_ · 2026-10-26
- Persistent AI joins UR network to route private AI inference with encrypted, IP-hidden traffic — 0xSammy · 2026-10-26
- Anthropic's disclosed compute commitments: $114.5B Google, $161.2B Broadcom top the list — BenBajarin · 2026-10-26
- SpaceX strikes $40B deal with Apollo Global to lease AI chips and data center hardware — Polymarket · 2026-10-26
- SpaceX AI revenue up 830%, Claude designs 318K antivenom proteins in 64 hours — PeterDiamandis · 2026-10-26