AI Radar

Codestral Mamba

Available · Language, Code

Codestral Mamba is a 7B code model from Mistral AI based on the Mamba architecture rather than a Transformer, which Mistral says gives linear-time inference. It was released under Apache 2.0 with weights on Hugging Face and offered for testing on la Plateforme; Mistral says it tested in-context retrieval up to 256k tokens. [1] [2] [3] Primary source

Timeline of Codestral Mamba →

Claims and evidence

  • Released Primary source[1]page date; last paragraph [4]July 16, 2024 [2]release date
  • Deprecated Primary source[5]Deprecated & retired models table, row Codestral Mamba 7B
  • Retired Primary source[5]Deprecated & retired models table, row Codestral Mamba 7B [2]Retirement date
  • Status AvailableApache 2.0 weights remain downloadable from the mistralai Hugging Face organisation; the API endpoint open-codestral-mamba was deprecated and retired on 2025-06-06. Primary source[3] [2]Weights table [5]Deprecated & retired models table, row Codestral Mamba 7B
      • Input text Primary source[2]Modalities
      • Output text Primary source[2]Modalities
      • Open weights YesApache 2.0. Primary source[1]fourth and fifth paragraphs [3]
      • Context window 256k tokensThe announcement says Mistral tested in-context retrieval up to 256k tokens. Primary source[2]Context; Weights table
      • Parameters 7,285,403,648 Primary source[1]fourth paragraph [2]Weights table
      • Access API, Open-weights download Primary source[1]fourth and fifth paragraphs

      Lineage

      Predecessors

      No known predecessor.

      Successors

      No known successor.

      Based on

      Not derived from another model.

      Variants and derived

      None recorded.

      Siblings

      None recorded.

      All ancestors

      None.

      All descendants

      None.

      Variants

      No variants recorded in this record.

      Related AI Radar coverage

      AI Radar coverage starts in June 2026; no coverage linked yet.

      All model releases from Mistral AI on AI Radar →