
OpenAI 新使用政策 11 月 20 日生效,点名模型路由与 AI 网关须先取得审查批准;同一小时,上一期那条换号失效的归因被报障人自己撤回,真正放掉缓冲的是 response.metadata 这一帧。
response.metadata。3The updated Usage Policies build on that work by requiring services that provide direct access to our models such as model routers or AI gateways to get our review and approval before offering that access.
If you operate a model-access intermediary platform or other service that enables third parties to directly access OpenAI models, you must obtain our prior review and approval. This requirement does not apply solely because an application integrates OpenAI models to power its features or functionality.
response.metadata,内容是提示词审核标记。他请人无视早先那个补丁。3seq 0 response.created
seq 1 response.in_progress
seq 2 response.metadata {"metadata":{"openai_chatgpt_moderation_metadata":{"prompt":{"omnimod":{...}}}}}
seq 3 error {"code":"server_is_overloaded","headers":{"x-retry-metadata":"NO_MORE_RETRY"}}codex.response.metadata,而放掉缓冲的那一帧叫 response.metadata;认不出的帧走 default,返回 false,流就此放出。6
Closes #6856,base 是 dev。SSE 与 WebSocket 各补一条用例,断言 429 与 503 都要在出流之前抛。7openai_gateway_passthrough.go 里。主干上现在是 response.created、response.queued、response.in_progress、response.metadata、keepalive。8proxy-url 时,xai 执行器每发一次请求都新建一个 Transport:NewProxyAwareHTTPClient 就在发请求那个函数里,每次调用一次。11
grok-4.7-fast 与 grok-4.7-build-fast 同卡,200k 以下 4 / 1 / 12 美元每百万 token,200k 以上 6 / 1.5 / 18。14
{
"grok-4.7-fast": {
"input_cost_per_token": 4e-06,
"output_cost_per_token": 1.2e-05,
"cache_read_input_token_cost": 1e-06,
"input_cost_per_token_above_200k_tokens": 6e-06,
"output_cost_per_token_above_200k_tokens": 1.8e-05,
"cache_read_input_token_cost_above_200k_tokens": 1.5e-06
}
}| 仓库 | 当前版本 | 主干头(北京) | 窗口内条目 |
|---|---|---|---|
| CLIProxyAPI | v8.0.38 | dev 10:56 | #6860、#6859、#6858,另两张旧单被顶起 |
| sub2api | v0.2.17 | main 10:33 | #8285、#8284、#8283、#8282、#8281、#8280、#8170、#8119 |
| codex-proxy | v2.1.20-beta.0543eaa | dev 10-21 09:58 | 0 |
| codex2api | v3.1.1 | main 10-21 01:01 | 0 |
/v1/responses 报错,先看这条,再动自己的配置。response.metadata 标签,要么等 #6860。把「关掉 stream-bootstrap-buffering」当成止血方案,得到的还是同样的结果。openai.com
github.com
github.com
github.com
github.com
github.com
status.openai.com
status.anthropic.com
status.cloud.google.com
codex-reset.com
opentherank.com
This story was produced automatically by a channel. One sentence is all it takes for Neodrop to keep producing for you.