{"data":{"HeaderNavModules":"{\"home\":true,\"console\":true,\"pricing\":{\"enabled\":true,\"requireAuth\":false},\"docs\":true,\"about\":true}","SidebarModulesAdmin":"{\"chat\":{\"enabled\":true,\"playground\":false,\"chat\":true},\"console\":{\"enabled\":true,\"detail\":true,\"token\":true,\"log\":true,\"midjourney\":true,\"task\":true},\"personal\":{\"enabled\":true,\"topup\":true,\"personal\":true},\"admin\":{\"enabled\":true,\"channel\":true,\"models\":true,\"deployment\":true,\"redemption\":true,\"user\":true,\"subscription\":true,\"setting\":true}}","announcements":[{"content":"如果你的 Codex App 或 CLI 里还看不到 GPT-6.2 Astra 和 Sol，打开[本站 Codex 模型目录](https://api.avemujica.moe/codex-catalog)，选一个版本，点「复制配置指令」发给 Codex。配置完成后，完全退出并重新打开 Codex，就能在模型选择器里切换。已有账号、密钥和服务商配置都会保留。\n\n日常使用推荐**基础版**，沿用官方上下文设置；需要更长上下文时，可选 **500K** 或 **1M**。这两个数字是客户端预设，实际可用长度取决于模型和接入渠道。\n\n本站每小时检查一次 OpenAI 官方目录。希望以后出新模型时本机也能自动跟进，可以在复制指令前勾选「在设备上每日同步」。","id":159,"publishDate":"2026-11-19T12:37:25+00:00","type":"success"},{"content":"gpt-6.2-sol、gpt-6.2-astra 已上线。按惯例，gpt-6.2-sol 与之前的 gpt-6.1-sol 一样，缓存读取按输入价格的 10% 计费；gpt-6.2-astra 暂时对齐官方定价，缓存读取为输入价格的 5%","publishDate":"2026-11-17T19:51:41.614915+00:00","type":"success"},{"content":"gpt-6.1-terra is now available in the promax group. As usual, cached pricing is 0.1","extra":"","id":158,"publishDate":"2026-10-26T18:56:47.750Z","type":"default"},{"content":"服务商于北京时间晚上七点对ceph存储池中的故障硬盘进行了切换，目前一切正常~~受服务商宿主机磁盘io问题影响，目前本站模型请求可能不时出现波动，带来的不便敬请谅解~~","extra":"","id":157,"publishDate":"2026-10-22T10:53:21.142Z","type":"default"},{"content":"The service provider began restarting at 3:00 AM Beijing time and completed maintenance around 3:13 AM. It probably won’t restart again (","extra":"","id":156,"publishDate":"2026-10-17T19:12:12.434Z","type":"default"},{"content":"# Announcement Regarding Security Vulnerability Patches and Brief Downtime in the KVM Virtualization Environment\n\nDear users,\n\nA severe security vulnerability affecting the KVM virtualization environment, **PwnEscape (CVE-2026-99999)**, was recently publicly disclosed. The vulnerability exists in the underlying Linux kernel network stack and could breach the security isolation between virtual machines and their host machines, putting host security at risk. [Vulnerability details](https://access.redhat.com/security/cve/cve-2026-99999)\n\nTo promptly eliminate this security risk, our infrastructure service provider will deploy security patches across the board and restart the affected host machines and virtual servers.\n\n**Maintenance window: October 17, 2026, 22:30 to October 19, 2026, 22:30 Beijing Time.**\n\nWhen maintenance involves this site's servers, the website and API services may experience a **brief interruption of approximately 5–15 minutes**, including failed access, failed requests, or disconnected connections. The time range above is the overall maintenance window; actual downtime is expected to account for only a small portion of it, with the specific timing arranged by the service provider.\n\nPlease save important content in advance and plan your usage accordingly. If the site is temporarily unavailable, please try again later. Thank you for your understanding and support.","extra":"","id":155,"publishDate":"2026-10-17T14:07:37.729Z","type":"default"},{"content":"After reviewing Codex quota usage, we found that gpt-6.1-sol consumes quota at a rate far higher than its actual model pricing, with the difference nearly double. To balance revenue and expenses, starting at 12:00 noon Beijing time on October 17, we will adjust the cache pricing for gpt-6.1-sol to match gpt-6-sol.","extra":"","id":154,"publishDate":"2026-10-15T03:43:04.639Z","type":"default"},{"content":"openai上了新风控，还在处理中","extra":"","id":153,"publishDate":"2026-10-14T22:08:42.375Z","type":"default"},{"content":"OpenAI has discontinued the gpt-5.5 model. We have disabled it; please migrate as soon as possible.","extra":"","id":152,"publishDate":"2026-10-14T15:58:41.535Z","type":"default"},{"content":"- The deepseek group is now available at 60% of the official price. 4.1-flash follows official off-peak and peak pricing (weekdays 9\\~12 and 14\\~18 are peak hours, when the price doubles); 4-pro receives a 40% discount based only on the official peak-hour price\n- The glm group is now available at 75% of the official price","extra":"","id":151,"publishDate":"2026-10-12T11:13:15.742Z","type":"default"},{"content":"近期观察到部分请求偶发该报错：status_code=502, response protection is unavailable\n部分用户的请求会持续触发，遇到该报错请尝试创建新会话，如果问题持续，可以附带相关使用环境（是否在codex中使用，相关上下文）发起工单，方便排查","extra":"","id":150,"publishDate":"2026-10-09T02:42:57.487Z","type":"default"},{"content":"Due to cost factors, the aws group multiplier has been increased to 0.3","extra":"","id":149,"publishDate":"2026-10-01T13:37:16.164Z","type":"default"},{"content":"gpt-6.1-sol is now available in the promax group","extra":"","id":148,"publishDate":"2026-09-29T19:12:06.078Z","type":"default"},{"content":"openai上新模型的时候疑似又炸了，目前可能不时出现波动","extra":"","id":147,"publishDate":"2026-09-29T17:55:07.317Z","type":"default"},{"content":"Forgot to mention, OpenAI was down around seven in the morning.","extra":"","id":146,"publishDate":"2026-09-26T08:55:21.250Z","type":"default"},{"content":"The Mid-Autumn Festival lottery is now live. Users who have been registered for at least 3 days can visit https://api.avemujica.moe/lottery to participate and win credits equivalent to 6/66/666 yuan, etc.","extra":"","id":145,"publishDate":"2026-09-24T11:34:46.904Z","type":"default"},{"content":"- gpt-6-luna is now available. Standard pricing uses the original price of gpt-5.6-luna. To reduce the risk-control burden, gpt-5.6-luna will be routed to gpt-6-luna, and gpt-5.6-luna will be completely discontinued later. Please migrate in advance.\n- As risk-control efforts remain ongoing, gpt-6-sol is currently available in the promax group. gpt-5.6-sol will also be discontinued later, so please migrate promptly.","extra":"","id":144,"publishDate":"2026-09-23T13:45:32.960Z","type":"default"},{"content":"成本有点兜不住了，暂时停止订阅购买，现有订阅仍然可用","extra":"","id":143,"publishDate":"2026-09-22T13:39:33.133Z","type":"default"},{"content":"grok-4.7已上线","extra":"","id":142,"publishDate":"2026-09-21T16:07:16.366Z","type":"default"},{"content":"openai风控改了，还在处理中，建议先用cc","extra":"","id":141,"publishDate":"2026-09-21T03:04:01.981Z","type":"default"},{"content":"重启完了家人们，一直等到五点才重启，希望炸的人不多（~~我们预计北京时间凌晨3点左右进行服务器升级，届时需要重启，预计会有几分钟的中断，带来的不便敬请谅解~~","extra":"","id":140,"publishDate":"2026-09-17T16:51:34.704Z","type":"success"},{"content":"做了些处理，暂时缓解了openai的新风控问题，本来想等openai不作妖后发一个抽奖，但好像等不到了，先半场开香槟发个抽奖活动吧，注册满三天用户访问链接即可参与：https://api.avemujica.moe/lottery 可获得1/10/100/500元额度\nterra目前报错率仍较高，建议优先sol/astra","extra":"","id":139,"publishDate":"2026-09-16T11:29:37.750Z","type":"default"},{"content":"openai从两点开始了大规模风控/服务异常，已经要炸完了，急用可用promax分组，但暂时可用率都不高，或者先用cc和kiro吧，kiro倍率已下调至0.2","extra":"","id":138,"publishDate":"2026-09-16T07:04:02.338Z","type":"default"},{"content":"1. Please do not use promax group distillation; distillation may result in account suspension with no refund.\n2. Please use the promax group only in codex.\n3. promax group status check: https://observatory.avemujica.moe/\n4. Currently on weekdays, OpenAI compute pressure is relatively high, and risk control and rate limiting are in effect.","extra":"","id":137,"publishDate":"2026-09-14T08:34:44.980Z","type":"default"},{"content":"gpt-image-2.5 is now available in the image generation group at the same price as gpt-image-2. It is currently uncertain whether the codex API actually uses 2.5, so you can test it yourself.\npromax group detection is now available: https://observatory.avemujica.moe/","extra":"","id":136,"publishDate":"2026-09-12T12:34:04.429Z","type":"default"},{"content":"Looks like it became 5.5 again and is easy to cause errors, let's switch back（","extra":"","id":135,"publishDate":"2026-09-10T18:44:39.086Z","type":"default"},{"content":"The default group currently can't handle concurrent requests, and all models except 5.5 are returning many errors. If urgent, you can use the promax group first; you will need to wait in the queue, but it is relatively available.","extra":"","id":134,"publishDate":"2026-09-10T08:44:34.398Z","type":"default"},{"content":"We have received word that the terra model is one of the reasons for risk control, and we will soon take terra offline. Please prepare to switch in advance.","extra":"","id":133,"publishDate":"2026-09-10T07:58:22.831Z","type":"default"},{"content":"Currently, the request error rates for openai's official gpt-6-astra and gpt5.6 series are relatively high. It is recommended to temporarily use gpt-5.5.","extra":"","id":132,"publishDate":"2026-09-10T03:48:35.262Z","type":"default"},{"content":"sol,astra has enabled long-context billing. The promax group can use astra, while astra in other groups may be dynamically unavailable due to different OpenAI risk-control levels. If you encounter a reconnection error, you can try switching to the promax group.","extra":"","id":131,"publishDate":"2026-09-07T19:12:53.598Z","type":"default"},{"content":"受openai风控政策影响，目前的rpm成本下我们无力维护低级模型，明天起luna的价格将调整回降价前原价，并且luna，terra都会带有长上下文计费，请受影响用户提前做好准备，我们更想能稳定提供旗舰模型","extra":"","id":130,"publishDate":"2026-09-07T14:21:21.391Z","type":"default"},{"content":"We have observed that some accounts no longer have access to the gpt-5.5 model. openai is expected to deprecate the gpt-5.5 model soon. Users who are still using it, please migrate to another model as soon as possible. We will deprecate gpt-5.5 after all accounts no longer have access.","extra":"","id":129,"publishDate":"2026-09-07T13:02:31.805Z","type":"default"},{"content":"Currently, gpt-6-astra has a relatively high risk-control level and is frequently overloaded. It is recommended to switch to gpt-5.6-sol.","extra":"","id":128,"publishDate":"2026-09-07T09:34:18.127Z","type":"default"},{"content":"由于openai目前对中转行为的限制，当检测到账号并发异常时，gpt-6-astra会被严重降智，为了缓解该情况，我们近期会对模型进行分流，旗舰模型和低级模型的账号会分开，以尽量避免rpm过高带来降智，同时我们上线了0.5倍率的promax分组，只提供astra和sol，尽量保证该分组不会因官方对中转降智受到影响（目前降智严重，暂时不要使用），该分组成本较高，也注意不要过高rpm并发（请注意，该分组不可使用订阅），同时目前观察发现astra模型在codex中的实际计费高于api标准价格，有效额度实际只有约百分之六十，后续可能也会按照实际消耗调整价格","extra":"","id":127,"publishDate":"2026-09-06T12:46:32.658Z","type":"default"},{"content":"The official OpenAI gpt-6-astra inference service seems to be down. You can switch to sol for now.","extra":"","id":126,"publishDate":"2026-09-05T14:55:36.568Z","type":"default"},{"content":"OpenAI has officially taken GPT-5.4 offline from the Codex channel. Please use another model.","extra":"","id":125,"publishDate":"2026-09-05T03:34:01.413Z","type":"default"},{"content":"gpt-6-astra is now live. Please note that this model is benchmarked against fable, and its price is relatively high.\nAt the same time, openai has not fully rolled out this model yet, so it may not currently appear in codex. You can visit this website to copy prompts and send them to codex, and let codex help you configure it: https://soyorin.bond/","extra":"","id":124,"publishDate":"2026-09-04T18:53:48.183Z","type":"default"},{"content":"1. The Kiro per-request group has been shut down. Switching to the usage-based group or cc is recommended.\n2. The special-price GPT group is currently relatively unstable. We recommend choosing the default group first.","extra":"","id":123,"publishDate":"2026-09-04T07:14:02.093Z","type":"default"},{"content":"Restored~~openai messed things up again, gpt exploded across the entire internet, waiting for the official fix~~","extra":"","id":122,"publishDate":"2026-09-03T14:46:28.462Z","type":"default"},{"content":"claude-fable-5-1 is now available. This model is expensive, so please use it with caution.","extra":"","id":121,"publishDate":"2026-09-02T10:35:28.685Z","type":"default"},{"content":"The MaoMao emby registration codes in the Benefits Center have been restocked. The requirement has also been adjusted so that they can be claimed with a historical cumulative recharge of 100, and renewal codes are now available, which can be claimed with a monthly cumulative recharge of 100 (https://api.avemujica.moe/benefits)","extra":"","id":120,"publishDate":"2026-08-30T13:26:13.999Z","type":"default"},{"content":"Restored~~Domestic payment gateways are currently under attack. You can use stripe payment for now (Alipay supported); the minimum limit has been lowered to 50~~","extra":"","id":119,"publishDate":"2026-08-28T09:17:24.217Z","type":"default"},{"content":"Due to OpenAI's current strict RPM limits, please do not make any high-concurrency or probing requests with small token counts. If this is triggered frequently, we will limit your account's RPM.","extra":"","id":118,"publishDate":"2026-08-27T11:27:10.173Z","type":"default"},{"content":"Due to an upgrade to the openai risk-control system, we need to introduce additional proxy pool egresses for the account pool, which may cause a slight increase in latency. Thank you for your understanding.","extra":"","id":117,"publishDate":"2026-08-25T16:45:18.431Z","type":"default"},{"content":"The check-in quota pack model has been switched to x-preview-f","extra":"","id":116,"publishDate":"2026-08-25T15:54:00.040Z","type":"default"},{"content":"【How to Resolve 401 After Upgrading to Codex 0.149.0】\n\nIf you encounter API_KEY_REQUIRED / 401 Unauthorized after the upgrade, it is because the new version no longer allows custom Providers to automatically inherit authentication from auth.json when requires_openai_auth=false.\n\nSolution:\n\nManually modify the configuration file\nPlease modify the configuration file:\n\nmacOS: ~/.codex/config.toml\nWindows: %USERPROFILE%\\\\.codex\\\\config.toml\n\nFind this line:\nrequires_openai_auth = true (change the original false to true to use it normally)","extra":"","id":115,"publishDate":"2026-08-21T09:15:57.376Z","type":"default"},{"content":"The model for the check-in pack has been changed to `x-preview-f`~~The deepseek-v4-flash channel is temporarily offline while we look for an alternative. We have added access to muse-spark-1.2 to the check-in quota pack, so it can be used as a temporary replacement~~","extra":"","id":114,"publishDate":"2026-08-20T23:32:35.759Z","type":"default"},{"content":"It has now recovered~~OpenAI was malfunctioning, and most requests were returning this error. We need to wait for the official fix. The current mitigation is to start a new conversation, try passing the old conversation's id to Codex, and ask Codex to read the conversation information and continue: status_code=400, Invalid prompt: your prompt was flagged as potentially violating our usage policy. Please try again with a different prompt: https://platform.openai.com/docs/guides/reasoning~~","extra":"","id":113,"publishDate":"2026-08-20T16:17:32.617Z","type":"success"},{"content":"Recently, openai modified the cache hit rules, resulting in a decline in the current cache hit rate. We are trying to optimize the scheduling logic to better match the updated cache hit rules.","extra":"","id":112,"publishDate":"2026-08-19T15:01:10.973Z","type":"default"},{"content":"Standard response: if you think it’s laggy, it’s an openai issue: {\"error\":{\"type\":\"service_unavailable_error\",\"code\":\"server_is_overloaded\",\"message\":\"Our servers are currently overloaded. Please try again later.\",\"param\":null}}","extra":"","id":111,"publishDate":"2026-08-18T07:13:25.365Z","type":"default"},{"content":"The base prices for the deepseek series models have been aligned with the official latest pricing rules and are charged based on peak and off-peak hours. Please adjust your usage time accordingly.","extra":"","id":110,"publishDate":"2026-08-17T07:52:46.231Z","type":"default"},{"content":"The check-in event is now live. Check in daily to receive a deepseek-v4-flash subscription quota pack worth up to $2. Enable “Subscription Priority” to use it.","extra":"","id":109,"publishDate":"2026-08-15T11:44:30.438Z","type":"default"},{"content":"OpenAI has pulled the plug, and the special offer is gone~~A special-offer sol group is now available for a limited time, with a 0.05 multiplier. Available with a subscription; the effective multiplier after subscription is 0.033. It may be taken offline at any time, so use it while you can~~","extra":"","id":108,"publishDate":"2026-08-15T08:35:14.767Z","type":"default"},{"content":"Grok has been restored.","extra":"","id":107,"publishDate":"2026-08-13T07:43:50.199Z","type":"default"},{"content":"grok-4.6 is now available. Due to cost issues, the grok group multiplier has been increased to 0.4.","extra":"","id":106,"publishDate":"2026-08-12T19:22:40.644Z","type":"default"},{"content":"If you find the special-price group slow, subscribe and use the default group（","extra":"","id":105,"publishDate":"2026-08-10T07:04:26.421Z","type":"default"},{"content":"Due to a vulnerability in the billing system, some image-generation requests triggered in the previous image-generation workspace may not have been charged. The issue has now been fixed, and we will deduct the quota retroactively. If you have an available subscription, the deduction will be taken from your subscription quota first.","extra":"","id":104,"publishDate":"2026-08-09T06:17:10.491Z","type":"default"},{"content":"对于使用icloud给我们发送邮件的用户请注意，如果您空间满了，我们的邮件是发不过去的，建议无特殊情况直接使用站内工单。同时针对领取了emby权益，但接收不到tg验证码的用户，可以尝试https://play.google.com/store/apps/details?id=org.thunderdog.challegram 该客户端，但该客户端只支持安卓系统，如果您需要在ios设备登录tg，需要先使用安卓系统用该软件接收验证码，完成登录，在ios设备登录时即可直接从在线的客户端接收验证码，无需通过手机号接收","extra":"","id":103,"publishDate":"2026-08-05T17:20:52.177Z","type":"default"},{"content":"Please do not use this site's luna model for distillation. We may take measures including, but not limited to, limiting rpm, banning, and other actions.","extra":"","id":102,"publishDate":"2026-08-05T12:57:11.113Z","type":"default"},{"content":"The delay before the first character may suddenly become several dozen seconds from time to time. This is not an issue on our end, but rather a fluctuation in OpenAI's service. There is nothing we can do about it, and it will generally recover on its own after waiting for a while.","extra":"","id":101,"publishDate":"2026-08-04T14:31:18.010Z","type":"default"},{"content":"Prices are very likely to increase; whether they actually do will depend entirely on OpenAI’s next moves. We will only consider making adjustments if risk controls continue to tighten, and we will notify you at that time. Thank you for your understanding.","extra":"","id":100,"publishDate":"2026-08-01T13:26:56.782Z","type":"default"},{"content":"deepseek-v4-flash has been synced with the latest official stable release. However, due to the currently high load on the official ds service, the following error may occur from time to time: `status_code=503, Service is too busy. We advise users to temporarily switch to alternative LLM API service providers. ` Usually, waiting and retrying, or trying again after a short while, should resolve the issue.","extra":"","id":99,"publishDate":"2026-07-31T07:49:18.322Z","type":"default"},{"content":"luna and terra models have been synchronized with the latest official prices, with luna reduced by 80% and terra reduced by 20%","extra":"","id":98,"publishDate":"2026-07-31T05:18:50.178Z","type":"default"},{"content":"Due to an unexpected incident in services related to domain name resolution, our website experienced downtime for a period of time. We are very sorry for the inconvenience caused. We have launched an apology lottery event. Users who have completed 30 days of registration can visit https://api.avemujica.moe/lottery to participate in the draw and can win prizes of 10 or 100 yuan credit.","extra":"","id":97,"publishDate":"2026-07-30T18:38:58.882Z","type":"default"},{"content":"When applying for an invoice, if you really need a special VAT invoice, you can apply for it directly; our recommendation is only intended for users who do not need a special VAT invoice（","extra":"","id":96,"publishDate":"2026-07-28T09:22:36.201Z","type":"default"},{"content":"openai目前服务出现异常，可能有部分数据中心炸了，出现报错等待重试即可","extra":"","id":95,"publishDate":"2026-07-25T09:21:52.156Z","type":"default"},{"content":"claude-opus-5 is now available in the AWS and CC groups.","extra":"","id":94,"publishDate":"2026-07-25T05:21:55.967Z","type":"default"},{"content":"The 0.3x discounted domestic model (deepseek-v4-flash,mimo-v2.5) is now live. Users with total top-ups over 500, please go directly to the Benefits Center to claim and use the 0.12x group.","extra":"","id":93,"publishDate":"2026-07-24T19:06:44.896Z","type":"default"},{"content":"The entire internet is going offline today—use it while you can and treasure it.","extra":"","id":92,"publishDate":"2026-07-24T04:15:28.558Z","type":"default"},{"content":"Due to proxy pool cost issues, the special-offer benefit model group multiplier (deepseek-v4-flash, mimo-v2.5) has been increased to 0.12","extra":"","id":91,"publishDate":"2026-07-23T11:08:27.237Z","type":"default"},{"content":"All channels have been hit by openai today; use them carefully while you can, because it's hard to say whether they'll crash.","extra":"","id":90,"publishDate":"2026-07-21T05:07:05.987Z","type":"default"},{"content":"The account is about to be fully blocked. We are doing our best to recover it, and delays and user experience may fluctuate for now.","extra":"","id":89,"publishDate":"2026-07-20T17:30:06.218Z","type":"default"},{"content":"We are optimizing cluster scheduling, and the first character may fluctuate for now.","extra":"","id":88,"publishDate":"2026-07-18T15:59:35.551Z","type":"default"},{"content":"If you find that when using Codex the same conversation keeps timing out and retrying repeatedly, this is usually a context issue. It may be that the previous content does not comply with OpenAI's content policy, or that GPT is overthinking. In this case, you can try interrupting it and ask it not to keep thinking, but to execute the task directly.","extra":"","id":87,"publishDate":"2026-07-17T05:44:29.850Z","type":"default"},{"content":"The special offer model group has launched: hy3（腾讯混元3）","extra":"","id":86,"publishDate":"2026-07-16T11:26:18.857Z","type":"default"},{"content":"Regarding the issue with image generation in Codex: https://github.com/openai/codex/issues/32153 According to this issue, the current official app version has a bug regression. Before the official fix is released, you can use the skill mentioned in the issue: https://github.com/jianxunzhang1101/recover-codex-images to recover the images you generated and continue using them (you can directly copy this message and ask Codex to install, recover, and use it).","extra":"","id":85,"publishDate":"2026-07-13T18:10:56.383Z","type":"default"},{"content":"The gpt-5.4 model appears to have been taken down in batches. Users still using it are advised to migrate in advance, as it will probably become unavailable soon.","extra":"","id":84,"publishDate":"2026-07-13T07:43:55.959Z","type":"default"},{"content":"- Cache write billing issue: OpenAI is currently returning 0 cache write tokens for the codex channel, so there is no cache write billing for now.\n- Tiered pricing issue: Based on current messages, tiered pricing has not yet been implemented for the codex channel, and we have removed the extra charge for long context.","extra":"","id":83,"publishDate":"2026-07-12T08:29:17.715Z","type":"default"},{"content":"grok has recovered (maybe\n~~grok got hit by large-scale risk control and temporarily crashed~~","extra":"","id":82,"publishDate":"2026-07-12T05:33:03.808Z","type":"default"},{"content":"For users encountering this error, please try updating your codex version: status_code=500, Invalid Value: 'tools'. Function 'image_gen.imagegen' conflicts with a hosted tool in the same request.","extra":"","id":81,"publishDate":"2026-07-11T13:20:51.344Z","type":"default"},{"content":"When submitting a ticket, do not start by saying “unreasonable”; please directly describe the problem you are experiencing. Also, when using Codex++ or other third-party applications to access this site’s API, do not select the OpenAI-compatible chat endpoint; you should choose the Codex native responses endpoint.","extra":"","id":80,"publishDate":"2026-07-11T12:59:19.843Z","type":"default"},{"content":"For the current issue in the codex app where 5.6 series models cannot be selected directly through API configuration, a helpful user provided a patch script: https://soyorin.bond/codex-statsig-patch.zip After extracting it and following the readme, close codex and run the script, and it will automatically open codex and show the configuration options for the 5.6 series models (I heard that after installing the latest version of the codex app, restarting your computer can also make them visible).","extra":"","id":79,"publishDate":"2026-07-10T14:36:31.498Z","type":"default"},{"content":"Need to clarify: OpenAI charges for cache creation for the 5.6 series models. We currently have not calculated this part of the consumption, but it will be pushed on the next backend update, but it is likely not going to be pushed today (","extra":"","id":78,"publishDate":"2026-07-10T09:13:15.211Z","type":"default"},{"content":"gpt-5.6 is currently only an alias used on this site, pointing to gpt-5.6-sol. It is recommended to use gpt-5.6-sol directly unless there is a special case.","extra":"","id":77,"publishDate":"2026-07-10T08:48:49.514Z","type":"default"},{"content":"Reasoning intensity explanation: in fact, there is currently no ultra reasoning level. If you select ultra in codex, it will be mapped to max when requesting. This is purely OpenAI hype; ultra only enables codex client’s proprietary multiagent, and the model itself does not have this reasoning level (","extra":"","id":76,"publishDate":"2026-07-09T22:45:52.887Z","type":"default"},{"content":"Should be fixed\n~~The luna model has special fingerprint verification; it will go live after the repair is completed~~","extra":"","id":75,"publishDate":"2026-07-09T21:41:22.214Z","type":"default"},{"content":"A backend update will be needed shortly, and during that time there may be a brief period of network retry errors. Please wait for recovery.","extra":"","id":74,"publishDate":"2026-07-09T20:54:32.604Z","type":"default"},{"content":"The gpt-5.4 model will be taken offline this month. Users who are still using it are advised to prepare to migrate to the gpt-5.6 series.","extra":"","id":73,"publishDate":"2026-07-09T19:46:22.118Z","type":"default"},{"content":"It is currently recommended to prioritize using gpt-5.6-sol. In codex, use the one that supports 372k context. The other two OpenAI model tiers also seem not to have been fully rolled out yet. At the same time, the 5.6 series models have stricter cyber-related risk controls, so please do not attempt any cybersecurity-related content requests. If risk controls are triggered frequently, we may enforce a ban.","extra":"","id":72,"publishDate":"2026-07-09T19:20:10.262Z","type":"default"},{"content":"At the same time, when the new model was launched, OpenAI also applied risk control to the previously used channel method. In addition, there is currently a supply issue with the 5.6-sol model, so we need to use a Pro account as a backup guarantee. We have now synchronized the GPT default group and Pro group multiplier to 0.3. It may be adjusted later based on the cost of Pro accounts. If users are concerned about costs, they can first use the special-price group, but the special-price group cannot currently guarantee support for the full 5.6 series models. In the future, we may permanently merge the default group and Pro group to ensure service stability. The special-price group will continue to be retained, but stability cannot be guaranteed.","extra":"","id":71,"publishDate":"2026-07-09T17:50:22.379Z","type":"default"}],"announcements_enabled":true,"api_info":[{"color":"green","description":"默认","id":2,"route":"默认","url":"https://api.avemujica.moe"},{"color":"blue","description":"EdgeOne线路，生图可用","id":3,"route":"EdgeOne","url":"https://api-eo.avemujica.moe"},{"color":"blue","description":"无法直连主站可用","id":4,"route":"CN","url":"https://api-cn.avemujica.moe"}],"api_info_enabled":true,"benefits_enabled":true,"cap_api_endpoint":"","cap_widget_script_url":"","chats":[{"Cherry Studio":"cherrystudio://providers/api-keys?v=1\u0026data={cherryConfig}"},{"AionUI":"aionui://provider/add?v=1\u0026data={aionuiConfig}"},{"流畅阅读":"fluentread"},{"CC Switch":"ccswitch"},{"DeepChat":"deepchat://provider/install?v=1\u0026data={deepchatConfig}"},{"Lobe Chat 官方示例":"https://chat-preview.lobehub.com/?settings={\"keyVaults\":{\"openai\":{\"apiKey\":\"{key}\",\"baseURL\":\"{address}/v1\"}}}"},{"AI as Workspace":"https://aiaw.app/set-provider?provider={\"type\":\"openai\",\"settings\":{\"apiKey\":\"{key}\",\"baseURL\":\"{address}/v1\",\"compatibility\":\"strict\"}}"},{"AMA 问天":"ama://set-api-key?server={address}\u0026key={key}"},{"OpenCat":"opencat://team/join?domain={address}\u0026token={key}"},{"绘图":{"key_selection_rules":[{"groups":[],"id":"rule-1780936569998-kjvmuk","models":["gpt-image-2"]}],"url":"https://gpt-image-playground.cooksleep.dev?apiUrl=https://api-eo.avemujica.moe\u0026apiKey={key}\u0026model=gpt-image-2"}}],"checkin_enabled":true,"custom_currency_exchange_rate":1,"custom_currency_symbol":"¤","data_export_default_time":"hour","default_collapse_sidebar":false,"default_use_auto_group":true,"demo_site_enabled":false,"discord_client_id":"","discord_oauth":false,"display_in_currency":true,"docs_link":"https://docs.avemujica.moe","email_verification":true,"enable_batch_update":false,"enable_data_export":true,"enable_drawing":false,"enable_task":true,"faq":[{"answer":"## Pricing Explanation\n\nThis site uses a **1:1 top-up** rule, and the base prices of models are configured according to official prices.\n\nTherefore, the actual discount calculation methods for different models are as follows:\n\n- **International models**: The actual discount is approximately `multiplier ÷ 7`; the specific result may be affected by exchange rate fluctuations.\n- **Domestic models**: The actual discount is the multiplier set in the backend.\n\nExample:\n\n```text\nInternational model multiplier 3.5 ≈ actual 50% discount\nDomestic model multiplier 0.5 = actual 50% discount","id":1,"question":"How is the discount calculated?"},{"answer":"We recommend using it in **Codex** for the most complete and smooth experience.\n\nSetup instructions: [Codex Configuration Guide](https://docs.avemujica.moe/cli/3-codex)\n\nFor everyday use, we recommend installing the **Codex App**:\n\n[Download Codex App](https://get.microsoft.com/installer/download/9PLM9XGG6VKS?cid=website_cta_psi)\n\nIf you mainly use it for coding, you can also use **Codex CLI**:\n\n```bash\nnpm install -g @openai/codex\n```\n\nIf you prefer **Claude Code**, you can also configure the site’s API based on client support.\n\nFor users who work with multiple tools such as Codex, Claude Code, and Gemini CLI, or who do not want to maintain configuration files manually, we recommend using **CC Switch** for unified management. CC Switch supports centralized management of Provider, MCP, Skills, Prompt, and other configurations, and also makes it easier to switch between different model services.\n\nLearn more: [CC Switch Official Site](https://cc-switch.cc)","id":2,"question":"How to use?"},{"answer":"If you encounter API_KEY_REQUIRED / 401 Unauthorized after upgrading, it is because the new version no longer allows custom Providers to automatically inherit authentication from auth.json when requires_openai_auth=false.\n\nSolution:\n\nManually modify the configuration file\nPlease modify the configuration file:\n\nmacOS: ~/.codex/config.toml\nWindows: %USERPROFILE%\\\\.codex\\\\config.toml\n\nFind this line:\nrequires_openai_auth = true (change the original false to true to use it normally)","id":11,"question":"How to resolve 401 after upgrading Codex 0.149.0"},{"answer":"We are currently prioritizing the availability of gpt; availability for other channels is not guaranteed at this time.","id":3,"question":"Channel Supply Issues"},{"answer":"OpenAI officially has a hard context limit for gpt-5.5 on this channel. It is recommended not to let your context exceed 258k. If you need a long context, please use gpt-5.4.","id":4,"question":"Why do 500 errors keep occurring?"},{"answer":"For image generation, it is recommended to use api-eo.avemujica.moe as the endpoint. The main node is affected by CF's maximum timeout limit, so image generation stability cannot be guaranteed for the time being.","id":5,"question":"The image generation model keeps reporting errors?"},{"answer":"This site provides an official GPT forwarding service. Please do not use it for malicious cybersecurity activities.\n\nOpenAI’s official systems perform cybersecurity checks on sessions. If `cyber_policy` is returned, it may lead to slower responses, additional checks, or account restrictions.\n\n## Disclaimer\n\nThis site does not read or store conversations. It only uses event error codes or exception types returned by OpenAI to warn, rate limit, or block frequent abuse.\n\nReferences: [cyber_policy](https://github.com/openai/codex/blob/12bfb57139f169c1a1822d57a6fdb2a3d08e0c53/codex-rs/codex-api/src/sse/responses.rs#L529-L535), [slowdown notice](https://github.com/openai/codex/blob/12bfb57139f169c1a1822d57a6fdb2a3d08e0c53/codex-rs/tui/src/chatwidget.rs#L189-L195)","id":6,"question":"Anti-Abuse Notice"},{"answer":"If you need an electronic special VAT invoice, please contact support@avemujica.moe in advance and recharge via corporate bank transfer. Electronic special VAT invoices and ordinary invoices are supported. The invoice amount for a single issuance must be no less than 500. If you have any special requirements, please contact support@avemujica.moe in advance.","id":8,"question":"Invoice Issues"},{"answer":"Contact information\nabuse@avemujica.moe Abuse\nsupport@avemujica.moe Support tickets\nbusiness@avemujica.moe Business cooperation\nIf you need an electronic special VAT invoice, please contact support@avemujica.moe in advance and recharge via corporate bank transfer","id":7,"question":"Have more questions?"},{"answer":"When transferring, please use an on-chain transfer, and make sure the payment amount matches what is displayed on the order page.","id":9,"question":"Web3 payment not received?"},{"answer":"Regarding the current issue where the 5.6 series models cannot be directly selected through the API configuration in the codex app, an enthusiastic user has provided a patch script: https://soyorin.bond/codex-statsig-patch.zip After extracting it, follow the readme, close codex and run the script. It will automatically open codex, and you will see the configuration options for the 5.6 series models.","id":10,"question":"Why can't I see the 5.6 series models in my Codex?"},{"answer":"Due to Cloudflare's 120s limit on request idle time, and because GPT does not return any tokens while thinking, please try using:\nhttps://api-eo.avemujica.moe for requests.","id":12,"question":"Does the request keep timing out after around 2m without receiving any tokens?"}],"faq_enabled":true,"footer_html":"","github_client_id":"Ov23liPyNpEBM4WpW4CZ","github_oauth":true,"human_verification_provider":"turnstile","invoice_enabled":true,"linuxdo_client_id":"","linuxdo_minimum_trust_level":0,"linuxdo_oauth":false,"locale":"en","logo":"https://cd.avemujica.moe/favicon.svg","lottery_enabled":true,"mj_notify_enabled":false,"oidc_authorization_endpoint":"","oidc_client_id":"","oidc_enabled":false,"passkey_allow_insecure":false,"passkey_attachment":"","passkey_display_name":"AveMujica API","passkey_login":false,"passkey_origins":"https://api.avemujica.moe,https://api-cn.avemujica.moe","passkey_rp_id":"avemujica.moe","passkey_user_verification":"preferred","password_register_enabled":true,"price":1,"privacy_policy_enabled":false,"quota_display_type":"USD","quota_per_unit":500000,"register_enabled":true,"self_use_mode_enabled":false,"server_address":"https://api.avemujica.moe","setup":true,"start_time":1795111628,"stripe_unit_price":1,"support_ticket_enabled":true,"system_name":"AveMujica API","team_workspace_enabled":false,"telegram_bot_name":"","telegram_oauth":false,"theme":"default","turnstile_check":true,"turnstile_site_key":"0x4AAAAAADHy1VSlRMT-U2oD","uptime_kuma_enabled":true,"usd_exchange_rate":1,"user_agreement_enabled":false,"version":"v1.6.6","wechat_login":false,"wechat_qrcode":""},"message":"","success":true}