ModelRefs / Agentic Systems — Canonical Workflow
Agentic Systems — Canonical Workflow
Canonical Agentic Systems workflow: tool-using models, planners, memory stores, benchmarks and deployment patterns.
Overview
Agentic systems give a model the ability to plan, call tools, and execute multi-step tasks autonomously, pairing reasoning models with structured tool schemas, memory stores, and circuit breakers to convert language into action while keeping cost and reliability in budget, unlike a single-turn chat interaction.
Use this page to check whether your use case needs autonomous multi-step execution rather than a simpler single-turn workflow, which models and managed-container or self-hosted-cluster architectures are compatible, and which benchmark references — such as HumanEval and MMLU — speak to tool-use and planning reliability for candidate models.
Workflow fit here is provisional decision support based on goal, data, constraints, risk, and available evidence — it does not guarantee reliable autonomous execution. Test the candidate stack on representative multi-step tasks, including failure recovery, cost under retries, and tool-call error handling, before relying on it for production automation, and keep a human-approval gate for high-consequence actions.
Implementation profile
| Category | agentic-models |
|---|---|
| Implementation maturity | production |
| Evidence status | partial |
| Primary use cases | agents, enterprise-automation |
| Deployment options | managed-api, self-hosted |
| Architectures | managed-container, self-hosted-cluster |
Candidate models with published references
- BGE-M3
- GPT-5
- GPT-5 Mini
- Claude Opus 4
- Llama 4 Scout
- DeepSeek R1
- Mistral Large 2
- Command R+
- o3
- o4 Mini
- Text Embedding 3 Large
- Claude Sonnet 4
Coverage means the model is a candidate worth evaluating for this workflow, not a ranking or a recommendation. Models whose reference pages are still in review are omitted.
Benchmarks relevant to this workflow
miracl, mkqa, mldr, swe-bench, aider-polyglot, gpqa, aime-2025, tau-bench, browsecomp-long-context, longfact-concepts, terminal-bench, mmmu, mmlu-pro, livecodebench.
Relevance is a coverage signal from the canonical registry. Each benchmark only describes its own protocol and date, so confirm the harness matches your workload before treating a score as evidence.
Continue your research
Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Agentic Systems — Canonical Workflow.