1. Choosing Models by Workload2. A/B Testing and Rollout Gates
Choosing Models by Workload
๐ Model Selection Strategyโฑ 11 minโญ 85 XPโจ Hands-on lab
One Model Is Rarely Optimal
Bedrock hosts 100+ foundation models across Amazon, Anthropic, Meta, Mistral, DeepSeek, Moonshot AI (Kimi), MiniMax, and OpenAI. Production systems route requests to different model tiers based on risk and required quality:
Low-risk summarization/classification: small/fast models (e.g. Amazon Nova Micro/Lite, Claude Haiku family).
Complex reasoning and agentic coding: frontier-tier models (e.g. Claude Opus/Sonnet family, Amazon Nova Pro/Premier, OpenAI's frontier models via the Responses API).
Open-weight/self-hostable needs: DeepSeek, Meta Llama, Mistral, or OpenAI's open-weight GPT-OSS models.
Regulated decisions: strongest controls + audit trail, regardless of model tier.
โจ HANDS-ON LABBuild a Routing Matrix
โญ +150 XP
Map support, analytics, coding, and compliance tasks to model families.
1Create a workload-to-model matrix file.
2Add an escalation rule.
lab-sandbox โ simulated environment
INFINITY LAB SANDBOX v2.6 โ simulated shell
Type the command for the current objective. Helpers: "hint", "solution", "clear".
$
OBJECTIVE 1 / 2 โ type "hint" if stuck
๐งช Knowledge Check
Press 1-4 to select1 of 2
Why do mature teams use model routing?
Because one model is always best
To optimize quality-latency-cost across workloads
To avoid testing
To remove telemetry
Choosing Models by Workload Tutorial | Model Selection Strategy โ AWS Bedrock Academy