Unverified50% confidenceBenchmarkExact time
On the GSM8K benchmark, D3D reaches a speedup of 7.48x with Qwen3-32B as the target model.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
11/5/2026
First Seen
Valid until: 2/3/2027
Sources
Related Entities
Related Claims
UnverifiedIn evaluations across seven benchmarks, D3D achieves an average decoding speedup of 5.00x with Qwen3-8B as the target model.78% similarUnverifiedUnder the same settings, DFlash achieves average decoding speedups of 4.03x and 3.67x with Qwen3-8B and Qwen3-32B as target models, respectively.76% similarUnverified6.14x is the highest speedup measured on the specific model Qwen3-8B; the report does not provide complete curves across models or tasks, so it should be understood as a peak value under a specific configuration rather than a universal end-to-end throughput improvement.70% similarUnverifiedIn an FP16-precision 4K×4K×4K transposed matrix multiplication task, the M5 Ultra with Neural Accelerator enabled surpassed 107 TFLOPS, which is 3.51 times its own non-accelerated performance and 4.79 times that of the M3 Ultra.65% similarUnverifiedOn the Artificial Analysis benchmark, Qwen 3.8 approaches or even surpasses larger models such as GLM 5.2 Max, Kimi K3, and GPT 5.6.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1137904API
curl https://kongchang.com/api/v1/knowledge/claims/1137904MCP
get_claim(id=1137904)