Neve N0.5 Flash
A large language model released by Neve AI targeting programming and AI development use cases. It uses a 309B-parameter Mixture-of-Experts architecture with 15.5B active parameters, natively supports a one-million-token context window, and is open-sourced under the MIT license.
Timeline (last 90 days)
Neve AI released Neve N0.5 Flash, a MoE architecture with 309B parameters and 15.5B activated, natively supporting a 1-million-token context, with weights and inference code open-sourced under the MIT license.
Neve N0.5 Flash's official benchmarks show a Deep SWE 1.1 score of 67.8 and Terminal Bench 2.1 score of 86.7, with API pricing of 0.6 yuan per million input tokens and 2.6 yuan per million output tokens.
All Facts (2)
Neve N0.5 Flash's official benchmarks show a Deep SWE 1.1 score of 67.8 and Terminal Bench 2.1 score of 86.7, with API pricing of 0.6 yuan per million input tokens and 2.6 yuan per million output tokens.
50%UnverifiedNeve AI released Neve N0.5 Flash, a MoE architecture with 309B parameters and 15.5B activated, natively supporting a 1-million-token context, with weights and inference code open-sourced under the MIT license.
50%