@ai-sdk/typesafe-ai
An npm package that provides a type-safe wrapper on top of the AI SDK. It locks provider-related dependency versions to ensure consistency between type definitions and runtime behavior.
Core Facts
typesafe-ai is a package that provides a type-safe call wrapper on top of the AI SDK, and the matching dependency versions must be locked to ensure consistency between type definitions and runtime behavior.
The reasoning retention feature only takes effect on models that support reasoning output and does not affect regular conversational models.
@ai-sdk/zai released version 3.0.32, a patch update (Patch Changes) containing only two commits.
Timeline (last 90 days)
The maxOutputTokens parameter applies to experimental decision calls and is passed through to the language-model-driven decision process to constrain the number of output tokens.
For native decision providers, if the backend does not support either the maxOutputTokens or the reasoning setting, the system issues a warning rather than silently ignoring it.
The core commit of this update is 0c2a9de, which adds two parameters, maxOutputTokens and reasoning, to experimental decision calls.
The version number jump from 3.0.21 to 3.0.22 follows the patch position increment in semantic versioning, indicating no breaking changes and no new public API.
The release notes for this version do not mention any CVE or security issues.
This release includes 6 commits since the last release.
For patch-level updates, if the changeset description is too brief or is not included in the release process, the Release page may show only the version number and build artifacts, lacking detailed user-facing explanations.
@ai-sdk/zai released version 3.0.32, a patch update (Patch Changes) containing only two commits.
Version 3.0.19 adds an image header as well as a file-level typesafe.reference provider option, allowing images to be referenced in instructions and criteria.
Version 3.0.19 adds an experimental typesafe.longContext provider option; when enabled, the context limit can be extended to 262k tokens.
25 more timeline events
All Facts (20)
typesafe-ai is a package that provides a type-safe call wrapper on top of the AI SDK, and the matching dependency versions must be locked to ensure consistency between type definitions and runtime behavior.
75%VerifiedThe reasoning retention feature only takes effect on models that support reasoning output and does not affect regular conversational models.
75%Verified@ai-sdk/zai released version 3.0.32, a patch update (Patch Changes) containing only two commits.
80%Verified@ai-sdk/svelte released version 5.0.93.
90%UnverifiedThe maxOutputTokens parameter applies to experimental decision calls and is passed through to the language-model-driven decision process to constrain the number of output tokens.
50%UnverifiedThe core commit of this update is 0c2a9de, which adds two parameters, maxOutputTokens and reasoning, to experimental decision calls.
50%UnverifiedFor native decision providers, if the backend does not support either the maxOutputTokens or the reasoning setting, the system issues a warning rather than silently ignoring it.
50%UnverifiedThe release notes for this version do not mention any CVE or security issues.
50%UnverifiedThe version number jump from 3.0.21 to 3.0.22 follows the patch position increment in semantic versioning, indicating no breaking changes and no new public API.
50%UnverifiedThis release includes 6 commits since the last release.
50%UnverifiedFor patch-level updates, if the changeset description is too brief or is not included in the release process, the Release page may show only the version number and build artifacts, lacking detailed user-facing explanations.
50%UnverifiedThe decision model in @ai-sdk/typesafe-ai version 3.0.19 supports image input in PNG and JPEG formats, implemented via multipart upload.
50%UnverifiedVersion 3.0.19 adds an image header as well as a file-level typesafe.reference provider option, allowing images to be referenced in instructions and criteria.
50%UnverifiedVersion 3.0.19 adds an experimental typesafe.longContext provider option; when enabled, the context limit can be extended to 262k tokens.
50%Unverified对于使用 @ai-sdk/typesafe-ai 的项目，3.0.18 属于低风险升级；建议留意 lock 文件中 provider 相关包版本是否被一并提升，避免多版本共存导致的类型不匹配
50%Unverified3.0.18 版本没有新增 API、没有破坏性变更，也没有功能说明
50%Unverifiedtypesafe-ai 通过提升自身版本并绑定特定的 provider 版本下限，向生态广播一个已知可用的组合
50%UnverifiedThe update adds image input handling capabilities to OpenAI Decisions and the language model adapter.
50%UnverifiedThis update does not modify any business logic and only aligns with the version iterations of the underlying provider and utility libraries
50%UnverifiedIt is recommended to run a type check and test suite after upgrading, and to upgrade provider-utils and provider to their corresponding versions at the same time
50%