This is the asynchronous batch-serving variant of Mistral's Codestral 2508, a dense code-focused model (256K-token context, released end of July 2025 under Mistral's Non-Production License) specialized for fill-in-the-middle completion, code correction, and test generation. The batch endpoint processes requests asynchronously at a discount versus the standard endpoint, fitting large, non-interactive code-generation jobs rather than latency-sensitive use.
Taxa de custo
0.1x
Contexto
256K
Lançamento
1 de ago. de 2025
Entrada
TextFile
Saída
Text
Suporte
RaciocínioChamada de ferramentasSaídas estruturadas