NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
A base class for all AI providers.
Last release today
20 Oct 2026
Release timing varies
gaps range from 2 months to 13 months
Nearly every release is documented
notes for 4 of 4 stable releases
Nothing withdrawn
no release was ever pulled
1 years old
4 releases · first in 2025
One column per month.
Added ImageGenerator (image generation: generateImage, getImageModels) and TextEmbedder (text embeddings: embedBatch, getEmbeddingModels, and embed, w
ImageGenerator (image generation: generateImage, getImageModels) and
TextEmbedder (text embeddings: embedBatch, getEmbeddingModels, and embed,
which splits large inputs into maxEmbeddingBatch-sized requests) mixins with
their request / response types.openai_compat.dart: OpenAI-format image / embedding body builders and
parsers, getOpenAiJson, and parseModelList.Add true SSE streaming support:
package:ai_box/provider_http.dart: postSseData() (streaming POST
with the same LLMException error normalization as requestJson())
and decodeSseJson().package:ai_box/openai_compat.dart: streamOpenAiCompletions() /
openAiChunksFromEvents() — incremental text / reasoning deltas and
cross-chunk tool-call assembly for OpenAI-compatible providers.LLMStreamChunk.reasoningDelta for incremental thinking output.
The final chunk now carries the complete parts of the response.LLMAIBase.chatStream(), the streaming counterpart of chat().http as a dependency.Remove seed from the base class.(Because claude does not support it)
- Initial release.
Your coding agent can read these notes before it upgrades. Set up the MCP server →