Unverified50% confidenceBenchmarkExact time
Jordan Ilchif tested Qwen 3.8 27B on a 32GB M4 MacBook Air, with Ollama's NVFP4 build running at 17.2 tokens/s, pure MLX at 6.6 tokens/s, and LLAMA.cpp at 4.4 tokens/s, but the formats were different.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/13/2026
First Seen
Valid until: 1/11/2027
Sources
Related Entities
Related Claims
UnverifiedRunning Gemma 4 26B on an M3 Max, MLX is 65.0 tokens/s and LLAMA.cpp is 63.9 tokens/s, essentially a tie between the two.73% similarUnverifiedIn a fair 4-bit comparison, MLX with D-Flash 2 speculative decoding ran at 16.6 tokens/s, versus Ollama Q4's 5.7 tokens/s.71% similarUnverifiedIn tests with the 4-bit quantized version (fitting entirely into 32 GB of VRAM), Ollama reached 240 tokens/s while FreeToken reached 225 tokens/s, making Ollama actually faster.70% similarUnverifiedBased on public data for high power plus MLX or Ollama, the decoding speed of different models on M4 Max is roughly a ladder as follows: Nemotron 3 Nano about 160 tokens/s, Qwen 3.5 35B at 107 to 130 tokens/s, Gemma 4 26B about 75 tokens/s, DeepSeek V4 Flash about 27 tokens/s, and dense 27B about 22 tokens/s.69% similarUnverifiedIn a llama.cpp environment with Qwen3 27B, the R9700 achieved 27.1 tokens/s (approximately 65% of bandwidth ceiling) and the B70 achieved 21.4 tokens/s (just over approximately 50%).68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1015152API
curl https://kongchang.com/api/v1/knowledge/claims/1015152MCP
get_claim(id=1015152)