Where ONTO fits — and where the alternatives are better.
Our take, in writing, on when to pick ONTO and when to pick something else. If you can find a place where this matrix isn't accurate, please open an issue.
The capability matrix.
Highlighted rows are where ONTO ships something the others don't have as a built-in primitive.
| Capability | Claude SDK | OpenAI SDK | Google ADK | ONTO |
|---|---|---|---|---|
| Default model | Claude 4.5/4.6 | OpenAI Responses API | Gemini | Ollama Cloud (free open-weight) |
| Memory model | Conversation buffer | Conversation buffer | Conversation buffer | Typed UPO + SMO |
| Plan mode | Feature inside Claude Code | Pattern via handoffs | Workflow-as-plan | First-class runtime mode |
| Languages | Python, TypeScript | Python | Python, TS, Go, Java | Python, TypeScript, Rust |
| Policy / consent | Tool allowlist | Guardrails | Tool allowlist | Per-call consent scopes |
| Ontology extraction | DIY | DIY | DIY | Native, 4-level |
| On-premise | No — hosted API | No — hosted API | GCP only | Yes — Helm + Postgres + Sled |
| MCP support | Yes — deepest ecosystem | Limited | Limited | Yes — stdio, HTTP, SSE |
| Claude Skills | Yes (canonical) | No | No | Yes — native SKILL.md |
| License | Anthropic Commercial | MIT | Apache-2.0 | MIT OR Apache-2.0 |
By competitor.
Same structure for each: what they win at, what ONTO wins at.
Claude Agent SDK
Claude Agent SDK wins
- Best for code-editing agents — Read/Write/Edit/Bash tools are state of the art.
- Deepest MCP ecosystem and hosted tracing.
- Claude is the strongest model for tool-use and long-context reasoning.
ONTO wins
- Typed durable memory across sessions (UPO + SMO), not just conversation buffers.
- Provider-agnostic — swap Claude, GPT, or open-weight via env var.
- On-prem ready: Sled/Postgres + local embeddings.
- Plan Mode is a first-class runtime mode with consent scopes.
- Rust SDK option.
OpenAI Agents SDK
OpenAI Agents SDK wins
- Best for voice and realtime agents.
- Tightest integration with OpenAI Responses API features.
ONTO wins
- Polyglot: native Python, TypeScript, and Rust SDKs from one Rust core.
- Memory model: typed UPO + SMO instead of conversation-as-memory.
- Cost meter for multi-tenant per-user/per-tenant billing.
- No vendor lock-in — open-weight default.
- Plan Mode + Policy Guard as runtime invariants.
Google ADK
Google ADK wins
- GCP-native; declarative workflow graphs.
- Go and Java SDKs available.
ONTO wins
- Conversational agent loop with durable memory (vs workflow graphs).
- Open-weight defaults; no GCP lock-in.
- On-prem ready: Helm + Postgres + local embeddings.
- Rust SDK for the highest-performance path.
LangChain / LangGraph
LangChain / LangGraph wins
- Largest ecosystem of integrations and community tutorials.
- LangGraph offers explicit graph workflow definition.
ONTO wins
- Typed memory model — not just conversation buffers or vector stores.
- Plan Mode is a first-class runtime invariant (not a graph node).
- Consent scopes ride on every tool call; not a separate audit layer.
- Polyglot core, Rust SDK option, polished cost meter.
Common comparison questions.
Why pick ONTO over the Claude Agent SDK?
The Claude Agent SDK is excellent for code-editing agents and has the deepest MCP ecosystem. ONTO wins where you need durable typed memory across sessions, multi-tenant scoping, Plan Mode as a first-class primitive, and the ability to run on open-weight models on-premise.
Why pick ONTO over the OpenAI Agents SDK?
The OpenAI Agents SDK is the right choice for voice and realtime use cases. ONTO is the right choice when you need a polyglot SDK (Python, TypeScript, and Rust from one core), durable typed memory, an honest cost meter for multi-tenant billing, and the freedom to swap providers without code changes.
Why pick ONTO over Google ADK?
Google ADK is GCP-native with declarative workflow graphs and Go/Java SDKs. ONTO is for teams that want a conversational agent loop with durable memory, open-weight defaults, and on-premise deployment freedom. Pick ADK if your team is committed to GCP and Go/Java; pick ONTO otherwise.
Why pick ONTO over LangChain or LlamaIndex?
LangChain is a chain orchestrator and LlamaIndex is a RAG framework. ONTO is an agent runtime with a typed memory model and explicit approval gates. If your problem is "retrieve documents and answer questions," LlamaIndex is the right tool. If your problem is "let an AI act on behalf of a user, with state and audit," ONTO is the right tool.
If ONTO fits, it's a one-liner to install.
If it doesn't, the matrix above will tell you what to pick instead — honestly.