提供 Claude API 中转服务，支持请求透传、按量计费和中国直连。
https://buzzai.cc主站端点
https://claude.buzz7.top推荐端点
完成站点认证，解锁以下权益：
https://acc.buzz7.tophttps://buzzai.top镜像端点
排名基于社区提交的测试数据与定期健康探测，仅供参考，非官方数据。
BUZZ 提供 Claude API 中转服务，将请求原样转发至 Anthropic 的 Claude 模型。支持 Claude Opus 4.6、Sonnet 4.6 和 Haiku 4.5 模型，确保请求和响应不被篡改—不替换模型、不裁剪上下文、不注入隐藏提示。服务采用多节点冗余保障可用性，原生支持 Prompt Cache 以优化 Claude Code 长会话成本，按量计费无月费或订阅。定价按百万 tokens 计算：Opus 4.6 输入 ¥0.7220，输出 ¥2.8880；Sonnet 4.6 输入 ¥0.1444，输出 ¥0.5776；Haiku 4.5 输入 ¥0.0361，输出 ¥0.1805。集成方式包括在 Anthropic 或 OpenAI SDK 中修改 base URL，Claude Code 用户可使用一键安装脚本。适用于 AI 开发、内容生成和数据处理等场景。
按 15 个模型行对比最新 audit、最新测速、吞吐、延迟与按 token 计费价格。
| 模型 | 输入 ($/M) | 输出 ($/M) | 检测 | 速度 | 延迟 |
|---|---|---|---|---|---|
gpt-5.6-lunaCodex-Pro | $0.060/M | $0.478/M | — | — | — |
gpt-5.6-terraCodex-Pro | $0.149/M | $1.19/M | — | — | — |
gpt-5.6-solCodex-Pro | $0.299/M | $2.39/M | — | — | — |
| $0.159/M | $0.797/M | — | — | — | |
claude-sonnet-5Corp Max | $0.378/M | $1.89/M | — | — | — |
| $0.438/M | $2.19/M | — | — | — | |
claude-fable-5Corp Max | $1.89/M | $9.46/M | — | — | — |
| $2.19/M | $10.95/M | — | — | — | |
| $0.398/M | $1.99/M | — | 50.5 t/s | 4.82 s | |
claude-opus-4-8Corp Max | $0.946/M | $4.73/M | — | 50.5 t/s | 4.82 s |
| - | - | — | 102.9 t/s | 9.88 s | |
qwen-opus:latest | - | - | — | 64.6 t/s | 7.59 s |
| - | - | — | 64.2 t/s | 3.55 s | |
| - | - | — | 137.0 t/s | 3.30 s | |
| - | - | — | 103.3 t/s | 3.30 s |
当前显示 15 / 15 个模型行
| 时间 | 模型 | 速度 | 延迟 |
|---|---|---|---|
| Jun 1, 09:19 AM | claude-opus-4-8 | 50.49 tok/s | 4.82s |
| Mar 27, 07:47 AM | qwen-opus:latest | 64.60 tok/s | 7.59s |
| Mar 23, 09:20 AM | gpt-5.1 | 140.43 tok/s | 3.06s |
| Mar 17, 10:21 AM | gpt-5.1 | 133.60 tok/s | 3.54s |
| Mar 17, 10:21 AM | gpt-5-codex | 103.25 tok/s | 3.30s |
| Mar 17, 10:19 AM | gpt-5.2 | 64.23 tok/s | 3.55s |
| Mar 16, 09:09 AM | grok-4.20-beta | 99.80 tok/s | 10.54s |
| Mar 16, 08:52 AM | grok-4.20-beta | 107.74 tok/s | 12.17s |
| Mar 16, 08:51 AM | grok-4.20-beta | 101.21 tok/s | 6.91s |
已修复 AWS 分组流式请求中断导致的异常多计费问题，6/28 起受影响用户已全额补偿至账户余额，详见日志「补偿/退款」记录。AWS 分组近期受上游风控影响可用性较低，建议改用 Corp Max / Codex-Pro 分组。 Fixed an issue where interrupted streaming requests on the AWS group were over-billed. Affected users (since Jun 28) have been fully compensated to their balance; see logs. AWS group availability is currently low due to upstream risk control — please use Corp Max / Codex-Pro instead. 已修復 AWS 分組串流請求中斷導致的異常多計費問題，6/28 起受影響用戶已全額補償至帳戶餘額，詳見日誌。AWS 分組近期受風控影響可用性較低，建議改用 Corp Max / Codex-Pro 分組。
New claude-sonnet-5 model added\n新增claude-sonnet-5模型
分组说明： - AWS分组：AWS逆向渠道。无降智 支持思考 支持努力程度等各项功能 缓存命中率高。 - MAX 分组：Anthropic 官方 Max 20x 订阅渠道，官方提示词优化、速度快、缓存命中率高。 - Claude-code-MAX：Anthropic 官方 Max 20x 订阅渠道，仅限Claude code中使用 （CLI VS插件）。 注： AWS 分组因渠道技术原因会影响模型身份认知（询问时回答其他名称），但模型本身真实有效，能力和体验不受影响
AWS逆向分组由于风控，可用性较低，请等待修复\nAWS reverse proxy group has low availability due to‑related risk‑control restrictions, please wait for repairs.\nAWS逆向分組由於風控，可用性較低，請等待修復
邀请返佣（AFF）正式上线！邀请好友充值即可获得返佣，可直接抵扣消费。同时，我们已向前期做出突出贡献的创始用户发放感谢补偿，余额可查。\nAffiliate Program (AFF) is now live! Invite friends and earn commissions on their top-ups, which can be used directly to offset your usage. We have also issued a thank-you bonus to our founding users for their early contributions. Please check your balance.
已修复流式请求断流仍计费的问题,6/2-6/12 受影响用户已全额补偿至账户余额,详见日志「异常断流计费补偿」。\nFixed an issue where interrupted streaming requests were still billed. Affected users (Jun 2-12) have been fully compensated to their balance; see logs.\n已修復串流請求中斷仍計費的問題,6/2-6/12 受影響用戶已全額補償至帳戶餘額,詳見日誌。
New claude-fable-5 model added\n新增cclaude-fable-5模型
AWS逆向渠道已降价，正在进行价格测试，若可行则会继续降价\nAWS reverse-channel pricing has been reduced and is currently undergoing price testing. If the results are positive, further price reductions will be implemented.
AWS逆向渠道恢复高水位全部缓存，测试阶段\nAWS reverse channel recovered high-water full cache, testing phase
New claude-opus-4-8 model added\n新增claude-opus-4-8模型
AWS逆向分组价格下调 0.04x，即刻生效。\n6/1 将进行新一轮价格调整，具体方案请关注后续公告。\n\nAWS reverse group pricing has been reduced by 0.04x, effective immediately.\nAnother pricing adjustment will take place on 6/1. Please follow upcoming announcements for details.\n\nAWS 逆向分組價格下調 0.04x，即刻生效。\n6/1 將進行新一輪價格調整，具體方案請關注後續公告。
AWS 逆向恢复\nAWS reverse service restored\nAWS 逆向恢復
AWS 逆向维护\nAWS reverse service maintenance\nAWS 逆向維護
新增 Codex-Pro 分组\nAdded Codex-Pro group\n新增 Codex-Pro 分組
AWS官方渠道下架\nAWS official channel removed\nAWS 官方渠道下架
5/20 SGT 起，本平台计价单位统一切换为 USD 锚定，分组倍率与账户余额同步换算，消费力不变。\nFrom 5/20 SGT, platform pricing switches to USD. Group rates & balances auto-converted — purchasing power unchanged.\n5/20 SGT 起，本平台計價單位統一切換為 USD 錨定，分組倍率與帳戶餘額同步換算，消費力不變。
新增AWS官方分组\nAdded AWS Official group\n新增 AWS 官方分組
AWS逆向上游资源不足 导致报错 重试即可\nAWS reverse upstream resources are insufficient, which may cause errors. Please retry.\nAWS 逆向上游資源不足，可能導致報錯，重試即可。
写代码优选 AWS 分组，模型无稀释，支持深度思考，缓存命中率 95%+，同等预算下产能更高。支持4.7\nAWS group recommended for coding: no model dilution, supports deep reasoning, cache hit rate 95%+, and delivers higher output under the same budget. Supports 4.7.\n寫程式優選 AWS 分組：模型無稀釋，支援深度思考，快取命中率 95%+，同等預算下產能更高。支援 4.7。
用 332 个 LMSpeed 信号，把 BUZZ 的 6 个相近 API 替代服务商放在一起比较：共享模型覆盖、价格、实测速度、可用性和免费模型。
| 服务商 | 对比理由 | 模型数 | 免费项 | 均价 | 速度 | 30 天可用性 |
|---|---|---|---|---|---|---|
| BUZZ buzzai-cc A Claude API proxy service offering transparent request forwarding, pay-as-you-go pricing, and direct connectivity in China. | 当前服务商基线 | 49 | 0 | $2.86/M | 96 tok/s | 89.3% |
| CaMeL AI api-kr777-top CaMeL AI provides an OpenAI-compatible API gateway with extensive model coverage and pricing options. |
| 194 | 4 | $88.30/M | 93 tok/s | 99.1% |
api-vectorengine-ai Vector Engine provides an API platform aggregating access to over 500 AI large models with OpenAI API compatibility and global deployment. |
| 182 | 6 | $5.75/M | 89 tok/s | 99.6% |
api-n1n-ai N1N provides API access to a wide range of AI models including GPT-4, Claude 3, Gemini, and others for text, image, and video generation. |
| 177 | 6 | $8.65/M | 90 tok/s | 99.7% |
api-futureppo-top Future Hub is a large-scale OpenAI-compatible API gateway at api.futureppo.top with extensive model coverage and pricing data. |
| 132 | 223 | N/A | 263 tok/s | 0.6% |
x666-me X666 API is an OpenAI-compatible API relay offering access to multiple AI models with competitive pricing and high availability. |
| 25 | 0 | $0.0014/M | 77 tok/s | 99.3% |
new-api-bxhm-onrender-com Kingo API Sharing Station is a LinuxDO API relay by user Kingo deployed on Render. Proxies Claude haiku / sonnet 4.5. Small daily sign-in credits; some 1.21-era models may return 502. |
| 3 | 0 | $1.71/M | N/A | 99.6% |
