v1.2.2

Highlights

• Timeouts for code that can hang: @tool(timeout=...) limits each function tool call, and timeout= on stream subscribers, @observer, BaseObserver, LoopDetector and TokenMonitor limits each callback. A timed-out tool returns a TimeoutError to the model, which can continue.
• Anthropic SDK 2.x support, plus new AnthropicConfig options: a typed speed and open_telemetry to configure or disable the SDK's own OpenTelemetry spans.
• OpenAIDecisionsConfig accepts question_instructions (strings, ag2 inputs or SDK parts), appended to the agent prompt.
• A2UI v1.0 updated to the current spec, including callAgentFunction / agentFunctionResponse and renderer capabilities.
• New Glasser paid-data toolkit extension.
• More accurate token usage: Anthropic continuation tokens, Gemini thinking tokens and xAI reasoning tokens are now counted, and OpenAI cache_write_tokens maps to cache_creation_input_tokens.
• Fixes for Mistral streaming, A2A empty text artifacts, eval dataset tags and failure attribution, SKILL.md files with a BOM, and update_file with empty old_content.

Upgrade notes

• OpenAIConfig is deprecated and becomes the Responses API in 1.4.0. The current Chat Completions behavior is now OpenAICompletionsConfig; OpenAIConfig still works unchanged but emits a DeprecationWarning from create(). Switch to OpenAICompletionsConfig to keep Chat Completions, or to OpenAIResponsesConfig to move now. Renaming the class does not translate API-specific options.
• temperature, top_p and top_k on AnthropicConfig are deprecated until 1.4.0. The API version used by anthropic 2.x rejects them, so with SDK 2.x they warn and are dropped; with 1.x they warn and are still sent. The extra now requires anthropic>=1.15.0,<3.
• OpenAI now requires openai>=3.37.0,<4.

What's Changed

• feat: add Glasser paid-data toolkit by @glasserai in #3462
• fix(mistral): isolate streamed completion choices by @ch-z-hc in #3471
• test(tools): keep the tight switch interval off under a tracer in the… by @vvlrff in #3481
• fix(anthropic): count continuation token usage by @ch-z-hc in #3472
• feat(a2ui): update A2UI v1.0 to the current spec by @vvlrff in #3469
• chore(deps): bump gitpython from 3.1.62 to 3.2.1 by @dependabot[bot] in #3485
• chore(deps): bump opentelemetry-api from 1.45.0 to 1.45.1 by @dependabot[bot] in #3486
• fix(a2a): preserve empty text artifact type by @fallintoplace in #3451
• fix(gemini): include thinking tokens in completion usage by @original4422 in #3446
• fix(filesystem): reject empty old_content in update_file on non-empty files by @BichengWang in #3303
• fix(textual): use correct optional import module by @fallintoplace in #3497
• fix(stream): preserve explicitly provided empty storage by @fallintoplace in #3498
• fix(openai): map cache_write_tokens to cache_creation_input_tokens by @rome-xi in #3499
• fix(eval): reject non-string dataset tags by @fallintoplace in #3496
• feat(anthropic): add typed speed option to AnthropicConfig by @its-amann in #3495
• fix(a2ui): keep v1.0 validation linear in nested function calls by @vvlrff in #3492
• fix(xai): include reasoning tokens in completion usage by @vvlrff in #3493
• test(openai): fix usage expectations after cache-write mapping by @Lancetnik in #3510
• chore(deps): bump the github-actions group with 2 updates by @dependabot[bot] in #3507
• fix(eval): check final response in failure attribution by @Jacqueline122 in #3506
• fix(skills): accept SKILL.md files that start with a UTF-8 BOM by @kabishou11 in #3502
• feat(anthropic): support anthropic SDK 2.x and deprecate sampling par… by @vvlrff in #3513
• chore(deps): bump certifi from 2026.7.22 to 2026.10.18 by @dependabot[bot] in #3515
• feat(tools): add per-invocation function tool timeouts by @Lancetnik in #3514
• feat(anthropic): expose the SDK's open_telemetry option on AnthropicConfig by @ninglinLiu in #3519
• chore: bump version to 1.2.2 by @Lancetnik in #3520
• feat(openai): structured question instructions on OpenAIDecisionsConfig by @yftx293 in #3500
• feat(stream): add subscriber and observer timeouts by @Lancetnik in #3518
• feat(openai): deprecate Chat Completions default ahead of 1.4.0 by @Lancetnik in #3517

New Contributors

• @glasserai made their first contribution in #3462
• @fallintoplace made their first contribution in #3451
• @original4422 made their first contribution in #3446
• @rome-xi made their first contribution in #3499
• @its-amann made their first contribution in #3495
• @Jacqueline122 made their first contribution in #3506
• @ninglinLiu made their first contribution in #3519
• @yftx293 made their first contribution in #3500

Full Changelog: v1.2.1...v1.2.2