List your company

Latency on S-NIAH (Long-context retrieval)

10.5Latency (s)Direct
3.09253.096103.1153.104Mar 20, 2026

Evaluation results

MethodLinks
DirectParadigm=P1: Direct, Backbone=Mistral, Model Size=7B2026.0310.5
DirectParadigm=P1: Direct, Backbone=Llama, Model Size=8B2026.0311.8
DirectParadigm=P1: Direct, Backbone=Qwen3, Model Size=8B2026.0312.3
DirectParadigm=P1: Direct, Backbone=Mistral, Model Size=Cdsrl2026.0316.8
DirectParadigm=P1: Direct, Backbone=Qwen3, Model Size=32B2026.0318.7
DirectParadigm=P1: Direct, Backbone=Mistral, Model Size=8x22B2026.0320.2
DirectParadigm=P1: Direct, Backbone=Llama, Model Size=70B2026.0321.4
λ-RLMParadigm=P3: λ-RLM, Backbone=Llama, Model Size=405B2026.0328.1
λ-RLMParadigm=P3: λ-RLM, Backbone=Qwen3, Model Size=235B2026.0331.4
λ-RLMParadigm=P3: λ-RLM, Backbone=Llama, Model Size=70B2026.0335.6
λ-RLMParadigm=P3: λ-RLM, Backbone=Mistral, Model Size=Cdsrl2026.0336.5
λ-RLMParadigm=P3: λ-RLM, Backbone=Qwen3, Model Size=32B2026.0338.2
λ-RLMParadigm=P3: λ-RLM, Backbone=Mistral, Model Size=8x22B2026.0341.8
DirectParadigm=P1: Direct, Backbone=Qwen3, Model Size=235B2026.0342.1
λ-RLMParadigm=P3: λ-RLM, Backbone=Qwen3, Model Size=8B2026.0345.9
DirectParadigm=P1: Direct, Backbone=Llama, Model Size=405B2026.0348.6
λ-RLMParadigm=P3: λ-RLM, Backbone=Llama, Model Size=8B2026.0348.7
λ-RLMParadigm=P3: λ-RLM, Backbone=Mistral, Model Size=7B2026.0352.3
RLMParadigm=P2: RLM, Backbone=Llama, Model Size=405B2026.0386.3
RLMParadigm=P2: RLM, Backbone=Qwen3, Model Size=235B2026.0398.6
RLMParadigm=P2: RLM, Backbone=Mistral, Model Size=Cdsrl2026.03120.4
RLMParadigm=P2: RLM, Backbone=Llama, Model Size=70B2026.03128.4
RLMParadigm=P2: RLM, Backbone=Qwen3, Model Size=32B2026.03142.8
RLMParadigm=P2: RLM, Backbone=Mistral, Model Size=8x22B2026.03155.1
RLMParadigm=P2: RLM, Backbone=Qwen3, Model Size=8B2026.03164.3
RLMParadigm=P2: RLM, Backbone=Llama, Model Size=8B2026.03178.2
RLMParadigm=P2: RLM, Backbone=Mistral, Model Size=7B2026.03195.7