Function Calling APIs
Provider Comparison函数调用 API
供应商对比

Decision brief决策摘要

TL;DR

Who this comparison is for这份对比适合谁

Use this guide if you are selecting a provider, migrating an existing agent, or building one orchestration layer across multiple models. It assumes you can make API requests, define JSON Schema, and execute tools in a trusted backend.

如果你正在选择模型供应商、迁移现有 Agent，或建设跨模型编排层，这份指南适合你。默认你已经能够调用 API、定义 JSON Schema，并在可信后端执行工具。

Start with one canonical tool contract先定义一份统一工具契约

The durable unit is not a provider payload. It is a provider-neutral definition, a normalized call, and a normalized result. Store the provider call ID separately so the adapter can reconstruct the required reply.

真正稳定的单元不是某家 Payload，而是供应商无关的工具定义、标准化调用与标准化结果。供应商调用 ID 单独保存，适配器才能重建正确的回传消息。

type ToolDefinition = {
  name: string;
  description: string;
  inputSchema: JsonSchema;
  sideEffect: "read" | "write";
};

type ToolCall = {
  provider: "openai" | "anthropic" | "google";
  providerCallId: string;
  name: string;
  arguments: unknown;
  raw: unknown;
};

type ToolResult = {
  providerCallId: string;
  ok: boolean;
  content: unknown;
  evidence?: { source: string; retrievedAt: string }[];
};

OpenAI vs Anthropic vs Google: protocol comparisonOpenAI、Anthropic 与 Google 协议对比

OpenAI implementation: correlate with call_idOpenAI 实现：用 call_id 关联结果

Send function tools with the request. When the model emits a function_call, parse its arguments, execute the approved tool, then append a function_call_output carrying the same call_id. Use tool_choice deliberately instead of forcing tools by default.

请求中发送函数工具。模型输出 function_call 后，解析参数、执行已授权工具，再用相同 call_id 回传 function_call_output。应有意识地配置 tool_choice，不要默认强制调用。

const first = await client.responses.create({
  model, input: userInput,
  tools: [{ type: "function", name: "get_quote", description,
    parameters: quoteSchema, strict: true }]
});

const call = first.output.find(x => x.type === "function_call");
const result = await executeSafely(call.name, JSON.parse(call.arguments));

const final = await client.responses.create({
  model, previous_response_id: first.id,
  input: [{ type: "function_call_output", call_id: call.call_id,
    output: JSON.stringify(result) }]
});

Anthropic implementation: preserve tool_use orderingAnthropic 实现：保持 tool_use 消息顺序

Claude returns one or more tool_use blocks, commonly with stop_reason: tool_use. The next user message must contain corresponding tool_result blocks. Put result blocks before any additional user text and mark execution failures with is_error: true.

Claude 会返回一个或多个 tool_use 块，通常伴随 stop_reason: tool_use。下一条 user 消息必须包含对应的 tool_result；结果块放在其他用户文字之前，执行失败用 is_error: true 标记。

const calls = response.content.filter(x => x.type === "tool_use");
const toolResults = await Promise.all(calls.map(async call => ({
  type: "tool_result",
  tool_use_id: call.id,
  content: JSON.stringify(await executeSafely(call.name, call.input))
})));

messages.push({ role: "assistant", content: response.content });
messages.push({ role: "user", content: toolResults });

Google implementation: preserve state and signaturesGoogle 实现：保留状态与签名

In the Interactions API, a function call is a step with id, name, and arguments; return a function_result referencing its call ID. Stateful continuation uses previous_interaction_id. In stateless flows, replay prior steps exactly, including required thought/signature metadata.

在 Interactions API 中，函数调用是包含 id、name、arguments 的 step；回传引用该调用 ID 的 function_result。有状态模式使用 previous_interaction_id，无状态模式则必须完整重放历史 step，包括所需的思考签名元数据。

const call = interaction.outputs.find(x => x.type === "function_call");
const result = await executeSafely(call.name, call.arguments);

const next = await client.interactions.create({
  model,
  previous_interaction_id: interaction.id,
  input: [{ type: "function_result", call_id: call.id,
    result }]
});

Build a thin provider adapter构建轻量供应商适配器

Each adapter should do four things only: serialize definitions, extract calls, serialize results, and preserve provider state. Validation, authorization, execution, retries, and evidence belong to shared infrastructure.

每个适配器只负责四件事：序列化工具定义、提取调用、序列化结果、保留供应商状态。校验、授权、执行、重试和证据应位于共享基础设施。

const turn = await adapter.ask({ messages, tools: canonicalTools });
const calls = adapter.extractCalls(turn);

const results = [];
for (const call of calls) {
  const args = validate(call.name, call.arguments);
  await authorize(identity, call.name, args);
  results.push(await executeIdempotently(call, args));
}

const answer = await adapter.continue({
  state: turn.state,
  toolResults: results
});

Schema design that survives all three APIs能跨三家工作的 Schema 设计

• Use small objects with explicit properties and predictable primitive types.
• Keep enums short and explain units, formats, time zones, and defaults in descriptions.
• Avoid deeply recursive schemas, ambiguous unions, and provider-specific extensions in the canonical layer.
• For OpenAI strict mode, objects need additionalProperties: false and required fields; represent optional values explicitly where needed.
• Validate again after decoding. Schema compliance does not prove a ticker exists, a date range is allowed, or a transfer is authorized.

• 优先使用小型对象、明确属性和可预测的基础类型。
• 枚举保持简短，并在描述里写清单位、格式、时区和默认值。
• 统一层避免深度递归 Schema、模糊 union 与供应商私有扩展。
• OpenAI 严格模式下，对象需要 additionalProperties: false 与 required 字段；可选值按需显式表达。
• 解码后再次校验。符合 Schema 不代表股票代码存在、日期范围被允许，或转账已经授权。

Tool choice, strict mode, and parallel calls工具选择、严格模式与并行调用

Streaming and multi-turn state流式处理与多轮状态

Never execute a tool from an incomplete argument fragment. Buffer deltas until the provider marks the call complete, decode once, validate once, and persist the provider call ID before execution. For stateless continuation, retain every provider-required message item and opaque signature exactly.

绝不能从未完成的参数分片直接执行工具。先缓存 delta，等待供应商确认调用结束，再统一解码、校验，并在执行前持久化调用 ID。无状态续接时，所有供应商要求的消息项与不透明签名都要原样保留。

Security, cost, and latency controls安全、成本与延迟控制

Common failure modes and fixes常见失败模式与修复

A provider-neutral validation benchmark供应商无关的验证基准

Do not publish a single “accuracy” score from a handful of prompts. Run the same versioned corpus against every provider and report success by failure class.

不要用少量 Prompt 得出一个笼统“准确率”。应对每家供应商运行同一份版本化测试集，并按失败类型分别报告。

Migration checklist between providers跨供应商迁移清单

1. Freeze a canonical tool registry and version every schema.
2. Map definition fields, tool-choice controls, call IDs, result envelopes, and conversation state.
3. Remove unsupported schema features before translation.
4. Replay production traces in shadow mode without side effects.
5. Compare call selection, argument validity, recovery, latency, and cost.
6. Canary read-only tools first; enable write tools only after idempotency and authorization tests pass.

1. 冻结统一工具注册表，并为每个 Schema 建立版本。
2. 映射工具字段、选择控制、调用 ID、结果信封和会话状态。
3. 翻译前移除目标接口不支持的 Schema 特性。
4. 以无副作用影子模式回放生产链路。
5. 比较工具选择、参数有效性、恢复、延迟和成本。
6. 先灰度只读工具；幂等与授权测试通过后再开放写工具。

QVeris implementation patternQVeris 实现模式

Provider-native function calling controls the model-facing message loop. MCP can standardize how a client discovers and invokes tool servers. QVeris sits at the capability layer: it helps an agent discover, inspect, route, call, and audit real-world tools and data, while your provider adapter preserves the required OpenAI, Anthropic, or Google protocol.

供应商原生 Function Calling 管理模型侧消息循环；MCP 可以标准化客户端发现和调用工具服务器的方式；QVeris 位于能力层，帮助 Agent 发现、检查、路由、调用并审计真实世界工具与数据，而供应商适配器继续保持 OpenAI、Anthropic 或 Google 所要求的协议。

FAQ

Official sources官方资料