Agent InfrastructureAutomated release watch

@ai-sdk/openai@4.0.78

@ai-sdk/openai v4.0.78 introduces support for reasoningEffortUpdate: 'none' for GPT-6 Sol and Luna, alongside validation for supported reasoning efforts.

Human read

Why this signal matters

In release @ai-sdk/openai@4.0.78 (commit 94d5d6d), support was added for setting reasoningEffortUpdate to 'none' for GPT-6 Sol and Luna models in both request-level options and positioned system messages. The update also includes validation logic that checks effort updates against the model's supported values: unsupported request-level updates trigger a warning and are omitted, while unsupported historical updates are rejected.

Agent parse

Actionable summary

@ai-sdk/openai updated to 4.0.78: Added reasoningEffortUpdate: 'none' for GPT-6 Sol and Luna across request-level options and positioned system messages. Enforced validation that warns/omits unsupported request-level updates and rejects unsupported historical updates.

Agent usefulness
77/100
Confidence
96%
Canonical data
JSON + Markdown
Next actions

What builders should check

  • Verify compatibility when using reasoningEffortUpdate with GPT-6 Sol and Luna models.
  • Ensure reasoning effort updates match the target model's supported efforts to prevent omitted request options or rejected historical updates.
Classification

Tags and routing

vercelai-sdktypescript
Related signals

Continue the thread