Unverified50% confidenceBenchmarkExact time
LFM2.5-24B-A2B outperforms Qwen3.6-35B-A3B, Gemma-4-26B-A4B-it, and Nemotron-3.5-Lightning (30B-A3B) on multiple instruction-following and agentic benchmarks, but lags on knowledge-intensive benchmarks such as SWE-Bench Verified and GPQA Diamond.
1
Sources
50%
Confidence
Short-term (~14 days)
Relevance
10/14/2026
First Seen
Valid until: 10/28/2026
Sources
Related Entities
Related Claims
UnverifiedOpen-source models such as LLaMA 3, Mistral, and Qwen 2.5 can already run smoothly on mainstream consumer-grade GPUs (such as the RTX 4090) at parameter scales ranging from 7B to 14B.63% similarUnverifiedAccording to NVIDIA's official tests, the optimal decoding combination for Qwen3 27B on Jetson is NVFP4 quantization combined with speculative decoding.63% similarUnverifiedDeepSeek-R1-Distill-Qwen-32B is based on Qwen-2.5-32B and was derived by distilling knowledge from the full 671B MoE architecture model.62% similarUnverifiedLFM2.5-24B-A2B is a pure-text reasoning model with no vision tower, always begins with a <think> block, and those needing image capabilities should use LFM2.5-VL-3B (the highest model in the LFM2.5 VL series).61% similarUnverified盘古2.0 Flash在GPQA Diamond科学推理测试中（Thinking模式）得分83.7，低于千问3.6 35B的86、27B稠密模型的87.8和DeepSeek V4 Flash的88.161% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1029965API
curl https://kongchang.com/api/v1/knowledge/claims/1029965MCP
get_claim(id=1029965)