跳到主要內容ATP Token

/v1/messages 的事件序列

/v1/messages 帶 stream: true 時，不論由哪個上游供應商服務，都會發出下方的事件序列。

請求#

在 /v1/messages 設 stream: true。

curl -N https://api.atptoken.ai/v1/messages \
  -H "Authorization: Bearer atp-..." \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-haiku-4-5",
    "max_tokens": 256,
    "stream": true,
    "messages": [{ "role": "user", "content": "Say hi in 3 words" }]
  }'

事件序列#

event: message_start
data: {"type":"message_start","message":{...}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

# one per delta:
event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"..."}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"end_turn"},"usage":{"output_tokens":N}}

event: message_stop
data: {"type":"message_stop"}

ping 事件可能出現在串流的任何位置（通常緊接在 message_start 和 content_block_start 之後），不帶任何資料，略過即可。

事件#

事件何時出現要讀什麼
message_start一次，最先message.id、message.model。這裡的用量不是最終值。
ping任何時候沒有內容，略過
content_block_start每個內容區塊一次index、content_block.type（text 或 tool_use）；工具區塊還有工具的 id 與名稱
content_block_delta每個區塊多次delta.text（text_delta）或 delta.partial_json（input_json_delta）
content_block_stop每個區塊一次index 這個區塊已完整
message_delta一次，接近結尾delta.stop_reason，以及帶最終 input_tokens、output_tokens 的 usage
message_stop一次，最後串流結束

回應#

claude-haiku-4-5 的完整回應：

event: message_start
data: {"message":{"id":"01M3F85T5YV4BEGMA014TVTC9E","type":"message","usage":{"input_tokens":32,"cache_read_input_tokens":0,"cache_creation_input_tokens":0,"cache_creation":{"ephemeral_1h_input_tokens":0,"ephemeral_5m_input_tokens":0},"output_tokens":1},"role":"assistant","content":[],"model":"claude-haiku-4-5-20251001","stop_sequence":null,"stop_reason":null},"type":"message_start"}

event: ping
data: {"type":"ping"}

event: content_block_start
data: {"index":0,"type":"content_block_start","content_block":{"type":"text","text":""}}

event: ping
data: {"type":"ping"}

event: content_block_delta
data: {"index":0,"type":"content_block_delta","delta":{"type":"text_delta","text":"Hi there"}}

event: content_block_delta
data: {"index":0,"type":"content_block_delta","delta":{"type":"text_delta","text":" friend!"}}

event: content_block_stop
data: {"index":0,"type":"content_block_stop"}

event: message_delta
data: {"type":"message_delta","usage":{"input_tokens":32,"cache_read_input_tokens":0,"cache_creation_input_tokens":0,"output_tokens":7},"delta":{"stop_sequence":null,"stop_reason":"end_turn"}}

event: message_stop
data: {"type":"message_stop"}

用量請讀 message_delta#

token 數一律從 message_delta 取，不要用 message_start。message_start 的用量不是最終值；最終的 input_tokens 與 output_tokens 在 message_delta。

Tool use#

模型呼叫工具時，區塊以 tool_use 開始、input 是空的，參數則以 JSON 片段分散在 input_json_delta 事件中。把同一個 index 的所有 partial_json 接起來，在 content_block_stop 時一次解析。第一個片段可能是空字串。

event: ping
data: {"type":"ping"}

event: content_block_start
data: {"index":0,"type":"content_block_start","content_block":{"id":"toolu_01PGo2evEeMN4XS8geMjN7Vs","input":{},"name":"get_weather","type":"tool_use"}}

event: ping
data: {"type":"ping"}

event: content_block_delta
data: {"index":0,"type":"content_block_delta","delta":{"type":"input_json_delta","partial_json":""}}

event: content_block_delta
data: {"index":0,"type":"content_block_delta","delta":{"type":"input_json_delta","partial_json":"{\"city\": "}}

event: content_block_delta
data: {"index":0,"type":"content_block_delta","delta":{"type":"input_json_delta","partial_json":"\"Taipe"}}

event: content_block_delta
data: {"index":0,"type":"content_block_delta","delta":{"type":"input_json_delta","partial_json":"i\"}"}}

event: content_block_stop
data: {"index":0,"type":"content_block_stop"}

event: message_delta
data: {"type":"message_delta","usage":{"input_tokens":672,"cache_read_input_tokens":0,"cache_creation_input_tokens":0,"output_tokens":39},"delta":{"stop_sequence":null,"stop_reason":"tool_use"}}

工具呼叫的 id（例如 toolu_…）請當成不透明字串，在 tool_result 裡以 tool_use_id 原樣送回。

停止原因#

值意義
end_turn模型回答完畢
max_tokens輸出達到 max_tokens，內容被截斷
tool_use模型要你執行工具並送回結果

stop_reason 依供應商對應：stop → end_turn，length → max_tokens。

錯誤#

串流開始前就發現的錯誤，會以一般 JSON 回應加上 HTTP 狀態碼回傳，不會有事件串流。開始讀事件前，請先檢查狀態碼（或 Content-Type: text/event-stream）。

401
沒帶金鑰或金鑰無效。回應內容是 `{"message":"API key has been revoked or does not exist.","error":"token_revoked"}`，沒有 `error` 物件，兩種格式都要處理。
402
`insufficient_quota`：專案餘額 ≤ 0。
403
`permission_denied`：這個專案沒開放該模型，或模型不存在。

如果連線在 message_stop 之前就中斷，請把回應視為不完整。重試建議見錯誤碼。

下一步

上一頁OpenAI SSE下一頁錯誤碼
這頁有幫助嗎？