Skip to main content
Migration recommendations in older entries reflect model availability at the time of the announcement. For current options, see the Shared model catalog and Dedicated model catalog.
Gemma 4 31B availability changes with Shared InferenceStarting September 3, 2026, gemma-4-31b is no longer available with Cerebras Shared Inference. This change doesn’t affect Dedicated Inference, where Gemma 4 31B remains available.Use qwen-3.8-27b for Shared Inference workloads. To continue using Gemma 4 31B, contact us to set up Dedicated Inference.
Deprecated zai-glm-4.7
Deprecated llama3.1-8b and qwen-3-235b-a22b-instruct-2507We recommend migrating to GPT OSS 120B.
Deprecating disable_reasoning parameter for zai-glm-4.7The disable_reasoning parameter will no longer be supported after July 21, 2026. Migrate to reasoning_effort="none" to disable reasoning on zai-glm-4.7.See the Reasoning guide and Chat Completions API reference for details.
Deprecated qwen-3-32b and llama-3.3-70bWe recommend migrating to GPT OSS 120B.
Migration Notes:
  • Qwen 3 32B had reasoning enabled by default. GPT OSS supports configurable reasoning via the reasoning_effort parameter. Review the Reasoning documentation before migrating.
  • Legacy strict mode (constrained decoding) behavior on llama-3.3-70b differs slightly from current models. Review the Structured Outputs and Tool Use documentation before migrating requests using tools and/or response_format in strict mode.
Deprecated zai-glm-4.6We recommend migrating to Z.ai GLM 4.7.
Deprecated qwen-3-235b-a22b-thinking-2507We recommend migrating to GPT OSS 120B.
Deprecated qwen-3-coder-480bWe recommend migrating to Z.ai GLM 4.7.
Deprecated llama-4-scout-17b-16e-instructWe recommend migrating to GPT OSS 120B or Llama 3.3 70B.
Deprecated llama-4-maverick-17b-128e-instructWe recommend migrating to GPT OSS 120B or Llama 3.3 70B.
Deprecated deepseek-r1-distill-llama-70bWe recommend migrating to Qwen 3 32B.
Deprecated qwen-3-235b-a22bWe recommend migrating to either Qwen 3 235B Instruct or Qwen 3 235B Thinking.
The llama3.1-70b model has been deprecated. We recommend transitioning to llama-3.3-70b, a more powerful model released by Meta.