Promptfoo v0.121.19 adds Bedrock GPT-5.6 and more providers
Decision Brief
What changedPromptfoo v0.121.19 adds support for Bedrock GPT-5.6, Meta Muse Spark, Open Interpreter, and xAI Grok 4.5, along with bug fixes.
Why it mattersNew providers let developers evaluate and red-team the latest models directly via Promptfoo without custom adapters.
Who should careAI coding tool users
Affected stackClaude CodeOpenAIxAI
Builder actionUpgrade Claude Code to v0.121.19
Source confidenceHigh · Official release / blog / repo
The update adds provider support for Bedrock GPT-5.6 frontier and GA, Meta Muse Spark, Open Interpreter, and xAI Grok 4.5. Claude Code credentials now respect the CLAUDE_CONFIG_DIR env var. New optional token-usage fields and helper functions are added, along with a goblin tactic for red teaming. Bug fixes include allowing 0 as a containment value, preventing Azure crashes on empty choices, and handling xAI response edge cases.
Summary basis: official / RSS sourceCompiled from the source scope noted above; the original remains authoritative.
Sources
- promptfoo（评测 / 红队）
LLM eval and red-teaming toolkit — releases covering safety, jailbreak, and guardrail testing.
- promptfoo（评测 / 红队）
留言
登入后即可留言，和其他 builder 交换实测心得。
还没有留言，抢头香。