chinese-ai-free-api
Skill0 starsChanged 3 months agoMITView on GitHub
国产大模型网页端逆向 API 方案:Cookie/Token 抓取 → 转 OpenAI 协议 → 接入 Agent。覆盖即梦/Seedance、Kimi、豆包、通义千问、智谱、DeepSeek 等十几个服务。
- Pipes a download into a shell
In this repository219 other files it gives its agents.
- agent-doctrineSkill
- agent-evaluation-frameworkSkill
- agent-visualization-flintSkill
- hermes-agentSkill
- bailian-cliSkill
- edge-browser-automationSkill
- computer-useSkill
- jupyter-live-kernelSkill
- backup-strategiesSkill
- browser-automationSkill
- cicd-pipeline-templatesSkill
- copybaraSkill
- devops-toolkitSkill
- email-deliverabilitySkill
- monitoring-observabilitySkill
- obsidian-maintenanceSkill
- obsidian-troubleshootingSkill
- openclaw-windowsSkill
- openclaw-workerSkill
- webhook-subscriptionsSkill
- windows-dev-environmentSkill
- windows-masterSkill
- zero-downtime-deploySkill
- codebase-inspectionSkill
- github-workflowSkill
- adaptive-communicationSkill
- agent-cost-dashboardSkill
- agent-evolutionSkill
- agent-governanceSkill
- agent-osSkill
- context-efficient-toolsSkill
- context-loss-awarenessSkill
- creative-synthesisSkill
- hermes-desktop-pluginsSkill
- emotional-intelligenceSkill
- graph-engineering-workflowSkill
- hermes-maintenanceSkill
- hermes-voiceSkill
- hermes-hygieneSkill
- identity-coreSkill
- income-loopSkill
- master-synthesisSkill
- meta-cognitionSkill
- reasoning-engineSkill
- self-preservationSkill
- teaching-abilitySkill
- temporal-awarenessSkill
- hermes-themesSkill
- thinking-visualizationSkill
- user-modelingSkill
- voice-full-restoreSkill
- knowledge-graph-traversalSkill
- mcp-browser-canvasSkill
- free-api-gatewaySkill
- native-mcpSkill
- web-reverse-apiSkill
- ai-edaSkill
- huggingface-hubSkill
- industrial-ai-siemens-mcpSkill
- industrial-ai-unifiedSkill
- The first 60 shown.
The file
Files
国产大模型网页端逆向 API
概述
国内几乎所有大模型服务都可以通过抓取网页端的 Cookie/Token 来调用内部 API,然后包装成 OpenAI 兼容格式,供 Agent 直接使用。LLM-Red-Team 团队已验证了十几种服务,原理完全一致。
通用三步法
- 抓 Token:登录目标网页 → F12 开发者工具 → 应用程序 → Cookies → 复制关键 cookie(通常是 sessionid / token / access_token)
- 找端点:在 Network 标签页执行一次操作(生成图片/发消息),找到 POST 请求,右键复制为 cURL
- 转 Python:用 Python requests/urllib 模拟,加 OpenAI 兼容包装层
已验证服务清单 (LLM-Red-Team 系列)
| 服务 | Cookie 名 | 免费额度 | 备注 |
| 即梦/Seedance | sessionid | 每日 66 分 | 支持图+视频 |
| Kimi | access_token | 每日有限次 | Moonshot |
| 豆包 | sessionid | 每日有限次 | 字节跳动 |
| 通义千问 | token | 每日有限次 | 阿里 |
| 智谱 GLM | token | 每日有限次 | 智谱清言 |
| DeepSeek | aws-waf-token(已变) | 每日有限次 | ⚠️ 2026.7 认证机制已改:旧 HWWAFSESID/HWWAFSESID/ds_session_id 全失效。Web 端只剩 aws-waf-token,API 端点 /api/v0/chat/completions 对所有无认证请求返回 HTML。认证 token 可能迁移至 localStorage,cookie-only 方案已不可用。 |
| 阶跃星辰 | token | 每日有限次 | Step |
| 海螺 AI | token | 每日有限次 | MiniMax |
| 讯飞星火 | token | 每日有限次 | Spark |
| 秘塔 AI | token | 每日有限次 | Metaso |
即梦/Seedance 专属方案
即梦有特殊之处:除了 LLM-Red-Team 的逆向方案,字节还提供了官方 CLI(dreamina)。
方案 A:逆向 API(已调通)
见 references/jimeng-reverse-details.md
关键点:
- 签名算法:MD5(
9e2c|{uri[-7:]}|{platform}|{version}|{timestamp}||11ac) - Cookie 格式含 sid_guard 等特殊字段
- 请求必须带公共参数:aid=513695, device_platform=web, region=CN
- 提示词放在 abilities.generate.core_param 深层嵌套结构中
- 图片端点:POST
/mweb/v1/aigc_draft/generate - 轮询端点:POST
/mweb/v1/get_history_by_ids
方案 B:官方 CLI(dreamina)
安装:curl -s https://jimeng.jianying.com/cli | bash
限制:
- 需要 VIP 会员,普通用户无法使用
- 通过 OAuth Device Flow 登录(二维码)
- 支持 text2image / image2video / Seedance 2.0 全系
- 版本 1.4.10(2026-06-26)
方案 C:Docker 一站式(jimeng-free-api-all)
仓库:zhizinan1997/jimeng-free-api-all(413⭐)
- Docker 一键部署,内置 Web 管理面板
- 支持所有图片模型(5.0-lite 到 2.0-pro)+ 所有 Seedance 视频模型
- 自动同步即梦官网模型列表
- 多路 Token 轮询
接入 Hermes 的方式
- 轻量 Python 脚本(推荐):每个服务一个脚本,几十行代码,本地
python xxx_api.py启动 HTTP 服务 - Docker 方案:拉 free-api 镜像,OpenAI 兼容端口暴露给 Hermes
- MCP Server:
wwwzhouhui/jimeng-mcp-server(74⭐)可直接作为 Hermes MCP 后端 - 百炼 CLI(2026.7 新增):阿里云百炼官方 CLI(
bl),安装见windows-master技能的references/bailian-cli-setup.md。提供视觉/图像/视频/语音等 10+ 能力,有免费额度,是 DeepSeek 逆向失效后的可靠替代方案。
注意事项
- 逆向 API 不稳定,服务端随时可能改参数格式
- 禁止商用,仅供学习和自用
- 每天免费额度有限,多账号轮询可叠加
- Docker 镜像约 280MB,Python 方案更轻量
- 2026.7 DeepSeek 认证已变:cookie-only 方案失效,建议改用百炼 CLI 作为替代
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

