Hero AI Models

overview hero ai does not use a single model for every feature for default swimlane managed routing, inference uses anthropic claude on amazon bedrock , with a different primary model per feature hero ai native action defaults to a faster haiku tier for playbook automation, while hero ai companion , agents (plan generation, remediation, playbook builder, and component agents), and code agent default to sonnet tiers suited to longer reasoning tasks when swimlane manages bedrock routing, each feature has its own primary model and automatic fallbacks if a model is unavailable or throttled you can replace defaults with custom llm docid\ deoy k3jmcpdm c ikrq4 ( turbine cloud or turbine platform ) or with turbine platform installer settings use the sections below to see default models by feature, how fallbacks work, and how to check data residency when you pick models in the product or configure your own bedrock endpoint default swimlane models and fallbacks for standard turbine cloud and turbine platform deployments, swimlane routes each hero ai feature to a primary claude model on bedrock if that model is unavailable or throttled (rate limits), the platform switches to the next fallback tier in that row when one exists on transient timeouts , the platform may retry the same model before failing over feature default (primary) secondary fallback tertiary fallback hero ai native action docid\ rx5y8x2 bzqjoup2j 8xq claude haiku 4 5 claude haiku 3 5 claude haiku 3 hero ai companion https //docs swimlane com/hero ai companion claude sonnet 4 5 claude sonnet 4 claude sonnet 3 5 v2 agents (plan generation, remediation, playbook builder, component agents) claude sonnet 4 5 claude sonnet 4 agents share the same bedrock model chain as hero ai companion (companion chat, ai soc plan and remediation flows, text to playbook, and related agentic workflows) code agent (playbook and code generation agent) uses a two tier sonnet chain only (sonnet 4 5, then sonnet 4; no tertiary fallback) condition what happens model unavailability switch to the next fallback tier for that feature (if configured) request throttling / rate limits switch to the next fallback tier for that feature (if configured) timeout (transient) retry the same model up to the configured attempt limit, then fail over if applicable no further fallback tier return an error to the user exact foundation model version strings may change as swimlane and aws update bedrock offerings tier order and failover rules stay as in the table above swimlane does not use your content to train or fine tune these foundation models when defaults do not apply the tables above describe swimlane managed default routing your environment may use different models when override applies to details custom llm docid\ deoy k3jmcpdm c ikrq4 turbine cloud (litellm) and turbine platform (litellm or custom bedrock) account custom llm settings replace default bedrock routing with your proxy or bedrock inference profiles turbine platform installer turbine platform on premises environment variables (for example llm model id medium , swimlane heroaichat models bedrock model ) can override defaults confirm with your swimlane contact which settings apply in your cluster hero ai native action model selection playbooks that set an explicit model the native action can target a specific bedrock model where the product exposes model selection respect regional availability for that model to see which bedrock models you can use in each aws region, see the aws documentation regional availability in amazon bedrock https //docs aws amazon com/bedrock/latest/userguide/models region compatibility html use that reference when you select models in hero ai native action or configure custom bedrock inference profile ids, so you choose models offered in the regions where you require data to stay usage (tokens and credits) hero ai consumption appears in the hero ai dashboard https //docs swimlane com/hero ai dashboard under admin panel usage reporting (tokens and credits, where enabled for your account)