Sravnenie freimvorkov Agentic AI: LangGraph vs CrewAI vs OpenAI Agents SDK vs Claude Agent SDK

AI-agenty pereshli ot issledovatelskikh demo k proizvodstvennym sistemam. Ozhidaetsya, chto bolee 60% korporativnykh AI-prilozhenii budut vklyuchat agentnye komponenty k 2026 godu. No sozdanie agentov s nulya - upravlenie tsiklami instrumentov, sostoyaniem, pamyatyu, obrabotka oshibok i koordinatsiya neskolkikh agentov - eto slozhnaya zadacha. Imenno dlya etogo sushchestvuyut freimvorki.

V 2026 godu dominiruyut chetyre freimvorka: LangGraph, CrewAI, OpenAI Agents SDK i Claude Agent SDK. Kazhdyi iz nikh prinimaet printsipialino drugoi podkhod k odnoi i toi zhe probleme: dat LLM sposobnost rassuzhdatt, planirovat, ispolzovat instrumenty i sotrudnichat.

Obzor

LangGraph: Stroitel grafov

LangGraph modeliruet workflow agentov kak orientirovannye tsiklicheskie grafy. Vy opredelyaete uzly (funktsii, vypolnyayushchie rabotu) i ryobra (perekhody mezhdu nimi, optsionalno uslovnye). Sostoyanie protekaet cherez graf i sokhranitsya cherez checkpointing.

Eto naibolee yavnyi i kontroliruemyi freimvork - vy sami podklyuchaete kazhdyi shag.

Klyuchevye kontseptsii

• StateGraph: opredelenie grafa s tipizirovannnym sostoyaniem
• Nodes: funktsii Python, preobrazuyushchie sostoyanie
• Edges: svyazi mezhdu uzlami, mogut byt uslovnymi
• Checkpointing: vstroennaya postoyannost dlya dlitelnykh workflow

Primer koda

from langgraph.graph import StateGraph, MessagesState, START, END
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="gpt-4o")

def call_agent(state: MessagesState):
    response = llm.invoke(state["messages"])
    return {"messages": [response]}

def should_continue(state: MessagesState):
    last = state["messages"][-1]
    if last.tool_calls:
        return "tools"
    return END

def call_tools(state: MessagesState):
    # Execute tool calls and return results
    results = []
    for tool_call in state["messages"][-1].tool_calls:
        result = execute_tool(tool_call)
        results.append(result)
    return {"messages": results}

graph = StateGraph(MessagesState)
graph.add_node("agent", call_agent)
graph.add_node("tools", call_tools)
graph.add_edge(START, "agent")
graph.add_conditional_edges("agent", should_continue, {"tools": "tools", END: END})
graph.add_edge("tools", "agent")

app = graph.compile()
result = app.invoke({"messages": [{"role": "user", "content": "What's the weather?"}]})

Silnye storony

• Tonkii kontrol nad kazhdym shagom i perekhodom
• Vstroennyi checkpointing i human-in-the-loop
• Polnyi paritet s TypeScript
• Rabotaet s lyubym provaidderom LLM
• Luchshe vsego podkhodit dlya slozhnykh workflow s uslovnym vetvleniem i tsiklami

Slabye storony

• Krutaya krivaya obucheniya - nuzhno ponimat konteseptsii teorii grafov
• Mnogo koda dlya prostykh sluchaev - bazovyi agent trebuet bolshe boilerplate, chem drugie freimvorki
• Otladka potokov grafov mozhet byt slozhnoi bez LangSmith

Tseny

Open-source (MIT). LangSmith (upravlyaemaya platforma nablyudaemosti) imeet platnye tarify dlya proizvodstvennogo monitoringa.

CrewAI: Sobiratel komand

CrewAI ispolzuet chelovecheskuyu metaforu: vy sobiraaete komandu spetsializirovannykh agentov, kazhdyi s rolyu, tselyu i predystoriyei. Agenty sotrudnichayut nad zadachami s pomoshchyu instrumentov, koordiniruemye protsessom (posledovatelnym, ierarkhicheskim ili konsensusnym).

Dumaite ob etom kak o naime komandy, gde u kazhdogo chlena est opredelyonnaya dolzhnost i spetsializatsiya.

Klyuchevye kontseptsii

• Agent: persona s rolyu, tselyu, predystoriyei i instrumentami
• Task: zadanie s opisaniem, ozhidaemym rezultatom i naznachennym agentom
• Crew: gruppa agentov, rabotayushchikh vmeste
• Process: strategiya vypolneniya (posledovatelnaya, ierarkhicheskaya, konsensusnaya)
• Flow: uroven orkestratsii na osnove sobytii dlya svyazyvaniya neskolkikh komand

Primer koda

from crewai import Agent, Task, Crew, Process

researcher = Agent(
    role="Senior Research Analyst",
    goal="Find comprehensive data about the given topic",
    backstory="You have 10 years of experience in technology research. "
              "You are thorough and always verify facts from multiple sources.",
    tools=[web_search_tool],
    verbose=True,
)

writer = Agent(
    role="Technical Writer",
    goal="Create clear, engaging technical content",
    backstory="You write for a developer audience. "
              "Your articles are practical and include code examples.",
    tools=[file_tool],
    verbose=True,
)

research_task = Task(
    description="Research the latest developments in WebAssembly in 2026. "
                "Focus on WASI, Component Model, and production use cases.",
    expected_output="A structured research document with key findings and sources.",
    agent=researcher,
)

writing_task = Task(
    description="Write a blog post based on the research. "
                "Include code examples and Mermaid diagrams.",
    expected_output="A complete blog post in Markdown format.",
    agent=writer,
    context=[research_task],  # Writer receives researcher's output
)

crew = Crew(
    agents=[researcher, writer],
    tasks=[research_task, writing_task],
    process=Process.sequential,
    verbose=True,
)

result = crew.kickoff()
print(result.raw)

Silnye storony

• Intuitivnaya abstraktsiya na osnove rolei - legko ponyat
• Bolee 100 vstroennykh integratsii instrumentov
• Obshchaya pamyat mezhdu agentami (kratkosrochnaya, dolgosrochnaya, sushchnostei)
• Samoe bolshoe soobshchestvo (~40k zvyozd na GitHub)
• Ierarkhicheskii protsess s agentom-"menedzherom", kotoryi delegiruet i validdiruet

Slabye storony

• Menee tonkii kontrol, chem u LangGraph - vy opredelyaete roli, a ne tochnye puti vypolneniya
• Ierarkhicheskii protsess mozhet byt nepredskazuemym, kogda agenty ne soglasny
• Otladka razgovorov neskolkikh agentov slozhnee, chem potokov odnogo agenta

Tseny

Open-source yadro (besplatno). CrewAI Platform: 99 $/mesyats (Teams) do 120 tys. $/god (Enterprise). Tsena zavisit ot aktivnykh komand i ezhemesyachnykh vypolnenii.

OpenAI Agents SDK: Router

OpenAI Agents SDK (dukhovnyi naslednik Swarm) fokusirruetsya na peredachakh - agenty peredayut razgovory drugim spetsializirovannym agentam. Eto samyi minimalisticheskii freimvork: agenty, instrumenty, peredachi i guardrails. Vot i vsyo.

Klyuchevye kontseptsii

• Agent: model + instruktsii + instrumenty + peredachi
• Handoff: peredacha drugomu agentu (modeliruetsya kak instrument, kotoryi LLM mozhet vyzvat)
• Guardrail: validatsiya vkhoda/vykhoda, zapuskaemaya parallel no s agentom
• Runner: vypolnyaet tsikl agenta
• Tracing: vstroennaya nablyudaemost dlya vsekh vyzovov LLM, ispolzovaniya instrumentov i peredach

Primer koda

from agents import Agent, Runner, handoff, InputGuardrail, GuardrailFunctionOutput
from pydantic import BaseModel

class SafetyCheck(BaseModel):
    is_safe: bool
    reason: str

async def content_safety(ctx, agent, input_text):
    result = await Runner.run(
        Agent(name="Safety", instructions="Check if input is safe. No PII."),
        input_text,
        context=ctx,
    )
    output = SafetyCheck.model_validate_json(result.final_output)
    return GuardrailFunctionOutput(
        output_info=output, tripwire_triggered=not output.is_safe
    )

billing_agent = Agent(
    name="Billing Agent",
    instructions="You handle billing inquiries. Be precise with numbers.",
    tools=[lookup_invoice, process_payment],
)

refund_agent = Agent(
    name="Refund Agent",
    instructions="You process refund requests. Always verify the order first.",
    tools=[lookup_order, issue_refund],
)

triage_agent = Agent(
    name="Triage Agent",
    instructions="Route the customer to the right specialist. "
                 "Ask clarifying questions if needed.",
    handoffs=[billing_agent, refund_agent],
    input_guardrails=[InputGuardrail(guardrail_function=content_safety)],
)

result = await Runner.run(triage_agent, "I need a refund for order #4521")
print(result.final_output)
# The triage agent routes to refund_agent, which processes the refund

Silnye storony

• Chistyi pattern peredach - estestvennyi dlya workflow routinga/triazha
• Guardrails rabotayut parallelno s vypolneniem (fail-fast, ne blokiruyushchie)
• Vstroennaya panel tracing dlya otladki
• Nesmotrya na nazvanie, podderzhivaet modeli ne ot OpenAI
• Minimalnaya abstraktsiya - legko ponyat i rasshirit

Slabye storony

• Menee zreloe upravlenie sostoyaniem, chem u LangGraph
• Net vstroennoi postoyannosti ili checkpointinga
• Ekosistema stroronnikh instrumentov menshe
• Dizain, orientirovannyi na peredachi, mozhet ne podkhodit dlya kazhdoi arkhitektury

Tseny

Open-source (MIT). Vy platite za token za lyubuyu ispolzuemuyu model.

Claude Agent SDK: Razrabotchik

Claude Agent SDK prinimaet drugoi podkhod: vmesto opredeleniya workflow ili rolei vy dayote agentu nabor instrumentov i pozvolyaete emu samomu razobr atsya, kak vypolnit zadachu. On ispolzuet tot zhe avtonomnyi tsikl, kotoryi pitaet Claude Code - chitat, deistvovat, proveryat, iterirovat.

Klyuchevye kontseptsii

• query(): glavnaya tochka vkhoda, zapuskayushchaya tsikl agenta
• Vstroennye instrumenty: Read, Write, Edit, Bash, Glob, Grep, WebSearch, WebFetch
• Polzovatelskie instrumenty cherez MCP: opredelenie instrumentov kak in-process MCP serverov
• Sub-agents: spetsializirovannye agenty, kotorym roditel mozhet delegirovat
• Sessions: podderzhanie konteksta mezhdu neskolkimi vzaimodeistviyami

Primer koda

import { tool, createSdkMcpServer, query } from "@anthropic-ai/claude-agent-sdk";
import { z } from "zod";

const searchDocs = tool(
  "search_docs",
  "Search the internal documentation for relevant information",
  { query: z.string().describe("Search query") },
  async ({ query }) => {
    const results = await vectorStore.similaritySearch(query, 5);
    return {
      content: [{ type: "text", text: results.map(r => r.pageContent).join("\n\n") }],
    };
  }
);

const docsServer = createSdkMcpServer({
  name: "docs",
  version: "1.0.0",
  tools: [searchDocs],
});

for await (const message of query({
  prompt: "Find how authentication works in our system and write a summary",
  options: {
    mcpServers: { docs: docsServer },
    allowedTools: ["Read", "Glob", "Grep", "mcp__docs__search_docs"],
  },
})) {
  if (message.type === "result" && message.subtype === "success") {
    console.log(message.result);
  }
}

Silnye storony

• Pervoclassnaya integratsiya MCP - podklyuchaites k lyuboi ekosisteme MCP serverov
• Vstroennye instrumenty dlya failovykh operatsii, terminala i dostupa v internet
• Avtomaticheskoe szzhatie konteksta dlya bolshikh kodovykh baz
• Parallelnye sub-agents dlya slozhnykh zadach
• Tot zhe dvizhok, chto i Claude Code - proverennyi na realnykh workflow razrabotki

Slabye storony

• Tolko modeli Claude - net podderzhki neskolkikh provaidderov
• Bolee novyi freimvork s menshim soobshchestvom
• Trebuet sredy Node.js dazhe dlya Python SDK
• Menee yavnyi kontrol workflow po sravneniyu s LangGraph

Tseny

Open-source. Standartnye tarify za tokeny Claude API. Managed Agents (khosting versiya): 0,08 $ za chas sessii plyus stoimost tokenov.

Kogda vybrat kakoi

Vyberite LangGraph, esli:

• Vam nuzhenn tochnyy kontrol nad kazhdym shagom workflow
• Vash sluchai ispolzovaniya vklyuchaet slozhnuyu uslovnuyu logiku i tsikly
• Vy khotite vstroennuyu postoyannost i checkpointy human-in-the-loop
• Vam nuzhno ispolzovat neskolko provaidderov LLM v odnom workflow

Vyberite CrewAI, esli:

• Vy khotite intuitivnuyu abstraktsiyu na osnove rolei
• Vasha zadacha vklyuchaet neskolko agentov s raznymi spetsializatsiyami
• Vam nuzhno, chtoby agenty sotrudnichali i peredavali drug drugu kontekst
• Vy tsenite samoe bolshoe soobshchestvo i naibolshee kolichestvo vstroennykh integratsii

Vyberite OpenAI Agents SDK, esli:

• Vash osnovnoi pattern - marshrutizatsiya razgovorov k spetsialistam
• Vam nuzhny guardrails, validiruyushchie vkhod/vykhod parallelno
• Vy khotite maksimalno prostuyu abstraktsiyu s minimalnnym boilerplate
• Vstroennyi tracing i nablyudaemost vazhny

Vyberite Claude Agent SDK, esli:

• Vashi agenty dolzhny chitat, pisat i vypolnyat kod
• Vy khotite pervoclassnuyu integratsiyu s MCP serverami
• Vam nuzhny avtonomnye agenty, kotorye iteriruyut i samoispravlyayutsya
• Vy uzhe ispolzuete Claude i khotite naibolee glubokuyu integratsiyu

Mozhno li kombinirovat freimvorki?

Da. Rasprostranyonnyi pattern - ispolzovanie odnogo freimvorka dlya orkestratsii i drugogo dlya otdelnykh agentov:

• LangGraph dlya obshchego grafa workflow
• CrewAI dlya konkretnoigo uzla, trebuyushchego sotrudnichestva neskolkikh agentov
• Claude Agent SDK dlya podzadach, svyazannykh s kodirovaniem, cherez MCP
• OpenAI Agents SDK dlya triazha i routinga na storone klienta

Freimvorki ne isklyuchayut drug druga. Ispolzuite to, chto podkhodit dlya kazhdoi chasti vashei sistemy.

Zaklyuchenie

Kazhdyi freimvork delaet chetkyy vybor:

• LangGraph optimiziruet kontrol - vy reshaete kazhdyi perekhod
• CrewAI optimiziruet sotrudnichestvo - agenty rabotayut kak komanda
• OpenAI Agents SDK optimiziruet prostotu - minimalnaya abstraktsiya, chistye peredachi
• Claude Agent SDK optimiziruet avtonomiyu - daite emu instrumenty i pozvolte rabotat

Pravilnyi vybor zavisit ot vashego workflow, vashei komandy i vashego sushchestvuyushchego steka. Vyberite tot, kotoryi sootvetstvuet vashemu osnovnomu sluchayu ispolzovaniya, khorosho izuchite ego i podklyuchite drugie, kogda stoknyyotes s ikh silnoi storonoi.