One Key for GPT, Claude, Gemini and 300+ Models — OpenAI-Compatible API at tryallapi.com
What is TryAllAPI?
If you ship LLM features for a living, you already know the pain: one Key for OpenAI, another for Claude, another for Gemini, plus scattered SDKs, billing dashboards, and region headaches. TryAllAPI (also branded MomoAPI on-site) is a multi-model AI API gateway built for that exact mess.
You get one OpenAI-compatible base URL, one API Key, and access to GPT, Claude, Gemini, and 300+ other models—including DeepSeek, Qwen, Kimi, Grok, GLM, MiniMax, and more, with image/video-capable models in the catalog. Point your existing OpenAI SDK at TryAllAPI, change the model string when you want a different brain, and keep shipping.
- Product: tryallapi.com
- Docs: docs.tryallapi.com
- Base URL: https://tryallapi.com/v1
- GitHub (WorkBuddy integration guide): xianyu110/workbuddy-tryallapi
TryAllAPI is a third-party aggregator. It is not affiliated with OpenAI, Anthropic, or Google. Model names, availability, and prices change—always trust the live console and pricing page.
Why an OpenAI-compatible gateway helps
Most of the tooling ecosystem already speaks OpenAI’s HTTP shape: chat/completions, Authorization: Bearer …, JSON message arrays. That means:
1. Less glue code. LangChain, LiteLLM, Cursor-style custom providers, WorkBuddy custom models, and countless scripts already know how to talk to an OpenAI-compatible endpoint.
2. One Key to rule the catalog. Instead of managing N vendor keys and N billing accounts, you top up a USD balance once and route traffic by model name.
3. Easy A/B. Swap gpt-… for claude-… or gemini-… without rewriting your client—only the model id changes.
If your app already works with the official OpenAI SDK, switching is usually a one-line base_url (or baseURL) change.
How to switch base_url (keep your SDK)
1. Sign up and create a Key
1. Register at tryallapi.com.
2. Top up a small USD balance (from $1).
3. In the console, create an API token and store it securely (env var / secret manager—never commit keys).
2. Point the OpenAI SDK at TryAllAPI
Python
from openai import OpenAI
import os
client = OpenAI(
api_key=os.environ["TRYALLAPI_API_KEY"],
base_url="https://tryallapi.com/v1",
)
resp = client.chat.completions.create(
model="gemini-3.8-flash", # copy exact id from the model plaza
messages=[{"role": "user", "content": "Summarize OpenAI-compatible routing in one sentence."}],
)
print(resp.choices[0].message.content)
Node.js
import OpenAI from "openai";
const client = new OpenAI({
apiKey: process.env.TRYALLAPI_API_KEY,
baseURL: "https://tryallapi.com/v1",
});
const resp = await client.chat.completions.create({
model: "gpt-6-sol", // example — use a live name from the console
messages: [{ role: "user", content: "Say hello from TryAllAPI." }],
});
console.log(resp.choices[0].message.content);
cURL
curl https://tryallapi.com/v1/chat/completions \
-H "Authorization: Bearer $TRYALLAPI_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "claude-sonnet-5-5",
"messages": [{"role": "user", "content": "Ping"}]
}'
3. Model names must match the plaza
Copy model ids exactly from the TryAllAPI model plaza / console (case and hyphens matter). Docs at docs.tryallapi.com walk through common models with call examples and troubleshooting.
Some clients want the full chat completions URL: https://tryallapi.com/v1/chat/completions. The OpenAI SDK usually only needs the /v1 base.
Pricing: pay-as-you-go USD from $1
Billing is prepaid, pay-as-you-go:
- Top up from about $1 (WeChat, Alipay, and cards are commonly supported on-site).
- Usage is deducted from your USD balance.
- Effective cost follows the platform rule: roughly official list price × group multiplier (exact math and multipliers are shown in the console).
- Larger top-ups may include discounts—check the top-up UI.
Always verify live numbers on tryallapi.com—rates move as upstream vendors change list prices.
Practical tip: keep a cheap, fast model as your default for drafting and batch jobs, and route hard coding or long-reasoning tasks to a stronger id only when quality matters. Same Key, different model string.
Who it’s for
TryAllAPI fits developers and small teams who:
- Already use the OpenAI SDK (or OpenAI-compatible proxies) and want Claude / Gemini / DeepSeek without rewriting clients.
- Need one balance instead of juggling multiple overseas cards and vendor dashboards.
- Build agents, chat UIs, RAG pipelines, or IDE plugins that should swap models often.
- Want a low-risk way to benchmark models on real traffic before locking into a single vendor.
It is not a replacement for vendor-specific advanced APIs when you need every proprietary feature of a single cloud.
Quick checklist before you go to prod
1. Env-only secrets — rotate any Key that ever appeared in a screenshot or chat.
2. Exact model ids — copy from the plaza; typos produce model errors.
3. Balance alerts — insufficient balance rejects requests; keep a small buffer.
4. Context limits — set max tokens to what the chosen model actually supports.
5. Disclaimer — third-party routing; not affiliated with OpenAI / Anthropic / Google; SLA and model list can change.
Get started
1. Open tryallapi.com and create an account.
2. Top up from $1 and create an API Key.
3. Set base_url / baseURL to https://tryallapi.com/v1.
4. Call /chat/completions with a model id from the plaza.
5. Read guides on docs.tryallapi.com.
One Key. OpenAI-compatible. GPT, Claude, Gemini, and 300+ models. Pay-as-you-go USD.
Built an AI zombie hug tool — aizombie.app: 2 photos → ~20s vertical clip, guest checkout $9.90
Site: aizombie.app
Turned the viral “Zombie Love Story” trend into a small paid product.
How it works:
1. Upload two different photos (couple / parent-child / you + cat or dog)
2. Pick a template (Love Story, Match Cut, Trend, Last Hug, Cat, Dog, …)
3. Get ~20s, 15 shots, 9:16 vertical video with sound — ready for TikTok / Reels / Shorts
One-time credits, no subscription required:
- Single $9.90 (one 720p render)
- Trio $24.90
- Creator $69.90
Guests can pay via Stripe — no account needed.
Stack: Seedance 2.5 + Cloudflare Workers.
Demo: aizombie.app
Templates: aizombie.app
How-to / why the two photos must be different people: aizombie.app
GitHub: xianyu110/aizombie
Feedback welcome — quality, checkout flow, or copy.
Mentions
One Key for GPT, Claude, Gemini and 300+ Models — OpenAI-Compatible API at tryallapi.com What is TryAllAPI? If you ship LLM features for a living, you already know the pain: one Key for OpenAI, another for Claude, another for Gemini, plus scattered SDKs, billing dashboards, and region headaches. TryAllAPI (also branded MomoAPI on-site) is a multi-model AI API gateway built for that exact mess. You get one OpenAI-compatible base URL, one API Key, and access to GPT, Claude, Gemini, and 300+ other models—including DeepSeek, Qwen, Kimi, Grok, GLM, MiniMax, and more, with image/video-capable models in the catalog. Point your existing OpenAI SDK at TryAllAPI, change the model string when you want a different brain, and keep shipping. - Product: tryallapi.com - Docs: docs.tryallapi.com - Base URL: https://tryallapi.com/v1 - GitHub (WorkBuddy integration guide): xianyu110/workbuddy-tryallapi TryAllAPI is a third-party aggregator. It is not affiliated with OpenAI, Anthropic, or Google. Model names, availability, and prices change—always trust the live console and pricing page. Why an OpenAI-compatible gateway helps Most of the tooling ecosystem already speaks OpenAI’s HTTP shape: chat/completions, Authorization: Bearer …, JSON message arrays. That means: 1. Less glue code. LangChain, LiteLLM, Cursor-style custom providers, WorkBuddy custom models, and countless scripts already know how to talk to an OpenAI-compatible endpoint. 2. One Key to rule the catalog. Instead of managing N vendor keys and N billing accounts, you top up a USD balance once and route traffic by model name. 3. Easy A/B. Swap gpt-… for claude-… or gemini-… without rewriting your client—only the model id changes. If your app already works with the official OpenAI SDK, switching is usually a one-line base_url (or baseURL) change. How to switch base_url (keep your SDK) 1. Sign up and create a Key 1. Register at tryallapi.com. 2. Top up a small USD balance (from $1). 3. In the console, create an API token and store it securely (env var / secret manager—never commit keys). 2. Point the OpenAI SDK at TryAllAPI Python from openai import OpenAI import os client = OpenAI( api_key=os.environ["TRYALLAPI_API_KEY"], base_url="https://tryallapi.com/v1", ) resp = client.chat.completions.create( model="gemini-3.8-flash", # copy exact id from the model plaza messages=[{"role": "user", "content": "Summarize OpenAI-compatible routing in one sentence."}], ) print(resp.choices[0].message.content) Node.js import OpenAI from "openai"; const client = new OpenAI({ apiKey: process.env.TRYALLAPI_API_KEY, baseURL: "https://tryallapi.com/v1", }); const resp = await client.chat.completions.create({ model: "gpt-6-sol", // example — use a live name from the console messages: [{ role: "user", content: "Say hello from TryAllAPI." }], }); console.log(resp.choices[0].message.content); cURL curl https://tryallapi.com/v1/chat/completions \ -H "Authorization: Bearer $TRYALLAPI_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-sonnet-5-5", "messages": [{"role": "user", "content": "Ping"}] }' 3. Model names must match the plaza Copy model ids exactly from the TryAllAPI model plaza / console (case and hyphens matter). Docs at docs.tryallapi.com walk through common models with call examples and troubleshooting. Some clients want the full chat completions URL: https://tryallapi.com/v1/chat/completions. The OpenAI SDK usually only needs the /v1 base. Pricing: pay-as-you-go USD from $1 Billing is prepaid, pay-as-you-go: - Top up from about $1 (WeChat, Alipay, and cards are commonly supported on-site). - Usage is deducted from your USD balance. - Effective cost follows the platform rule: roughly official list price × group multiplier (exact math and multipliers are shown in the console). - Larger top-ups may include discounts—check the top-up UI. Always verify live numbers on tryallapi.com—rates move as upstream vendors change list prices. Practical tip: keep a cheap, fast model as your default for drafting and batch jobs, and route hard coding or long-reasoning tasks to a stronger id only when quality matters. Same Key, different model string. Who it’s for TryAllAPI fits developers and small teams who: - Already use the OpenAI SDK (or OpenAI-compatible proxies) and want Claude / Gemini / DeepSeek without rewriting clients. - Need one balance instead of juggling multiple overseas cards and vendor dashboards. - Build agents, chat UIs, RAG pipelines, or IDE plugins that should swap models often. - Want a low-risk way to benchmark models on real traffic before locking into a single vendor. It is not a replacement for vendor-specific advanced APIs when you need every proprietary feature of a single cloud. Quick checklist before you go to prod 1. Env-only secrets — rotate any Key that ever appeared in a screenshot or chat. 2. Exact model ids — copy from the plaza; typos produce model errors. 3. Balance alerts — insufficient balance rejects requests; keep a small buffer. 4. Context limits — set max tokens to what the chosen model actually supports. 5. Disclaimer — third-party routing; not affiliated with OpenAI / Anthropic / Google; SLA and model list can change. Get started 1. Open tryallapi.com and create an account. 2. Top up from $1 and create an API Key. 3. Set base_url / baseURL to https://tryallapi.com/v1. 4. Call /chat/completions with a model id from the plaza. 5. Read guides on docs.tryallapi.com. One Key. OpenAI-compatible. GPT, Claude, Gemini, and 300+ models. Pay-as-you-go USD.
[分享] Awesome Seedance 2.5： 1024 条公开提示词+封面画廊，附 20 个可安装 Agent Skills 整理了一份 ByteDance Seedance 2.5 的社区案例库，只收「明确标注 2.5 + 公开提示词」的作品。 亮点：1024 条封面+可复制提示词； Pages 画廊； 20 个 Agent Skills 。 仓库：[github.com]( xianyu110/awesome-seedance-2.5) 画廊： [xianyu110.github.io]( xianyu110.github.io) 安装：npx skills add xianyu110/awesome-seedance-2.5
最新的生图我放下面了！ xianyu110/awesome-nano-banana
Built an AI zombie hug tool — aizombie.app: 2 photos → ~20s vertical clip, guest checkout $9.90 Site: aizombie.app Turned the viral “Zombie Love Story” trend into a small paid product. How it works: 1. Upload two different photos (couple / parent-child / you + cat or dog) 2. Pick a template (Love Story, Match Cut, Trend, Last Hug, Cat, Dog, …) 3. Get ~20s, 15 shots, 9:16 vertical video with sound — ready for TikTok / Reels / Shorts One-time credits, no subscription required: - Single $9.90 (one 720p render) - Trio $24.90 - Creator $69.90 Guests can pay via Stripe — no account needed. Stack: Seedance 2.5 + Cloudflare Workers. Demo: aizombie.app Templates: aizombie.app How-to / why the two photos must be different people: aizombie.app GitHub: xianyu110/aizombie Feedback welcome — quality, checkout flow, or copy.
[分享] Awesome Nano Banana 2.1 —— 精选图片案例 + 公开提示词（21 条） 整理了一个 Google Nano Banana 2.1 （ gemini-nano-banana-2.1 ）的案例合集，只收「有图 + 有公开提示词」的条目，方便直接复用。 目前 21 条，涵盖人像、对比、文字设计等。 GitHub： xianyu110/awesome-nano-banana-2.1 在线画廊： xianyu110.github.io 欢迎提 PR / Issue 补案例。
分享一个 Claude Opus 5.5 可视化案例精选仓库（114 精选 / 1019 全量 + 提示词） 刷 X 上 Opus 5.5 的游戏 / 3D / 动效 / 讲解视频太散了，这位作者整理了一份 awesome 清单： [github.com]( xianyu110/awesome-claude-opus-5.5) 从约 1019 条真实帖子里手挑 114 个最值得看的 分类：游戏、动效、Three.js 、Blender 、讲解动画、交互可视化、Agent 工作流、同题对比等 能找到的提示词也收了（有完整 prompt 可直接跑） 在线画廊可搜可筛可播： [xianyu110.github.io]( xianyu110.github.io) 顶流大概是：相机对焦实验室（ 1.6 万赞级）、动效开源模板（ 1.2 万）、五模型打桥、Splatoon 可玩版、Rust→汇编 17× 等。 做 AI demo / 找提示词复现的朋友可以收藏。欢迎补充 PR 。
开源了 12 个电商短视频 Agent Skills：一张商品图 → 带配音字幕的 15 秒竖屏视频，本地 ffmpeg 出片 之前发过一个电商出图的 Skills ，有人问能不能接着把图变成视频。这周把这部分也做完了，开源（ MIT ）： xianyu110/ecommerce-video-skills 想解决的问题：让大模型写带货分镜很容易，但写完还得自己开剪映一镜一镜拖。我想让 agent 不光写脚本，还能把片子真剪出来。 做法：让 agent 输出一份 storyboard.json （每镜：用哪张图、怎么动、说什么、字幕高亮哪个词），然后本地一条命令渲染： python scripts/assemble.py storyboard.json -o out/demo.mp4 脚本做的事：edge-tts 逐句配音 → 镜头时长按配音实际长度反推 → 静图做 Ken Burns （逐帧 scale+crop ，比 zoompan 稳）→ xfade 转场（居中在镜头分界，不吃配音）→ ASS 字幕 / 花字 / CTA → 合成一段免版权 BGM ，有人声时自动压低 → 两遍 loudnorm 到约 -14 LUFS → MP4 + SRT 。纯 Python + ffmpeg ，剪辑这一步不需要任何 API Key （ edge-tts 需要联网）。 效果：下面这条只用了一张 800×800 的随手拍，14.5 秒、1080×1920 ，我机器上 27 秒左右渲染完：  MP4： upload.maynor1024.live 拆成了 12 个 skill，每个文件夹一个 SKILL.md ，可以只装需要的： - 3 秒钩子（ 9 种类型，≤18 字，带广告法禁用词自检） - 卖点分镜（ 15/30/60 秒结构，直接输出 storyboard.json ） - 图生视频提示词 / 多模型适配（ Seedance 、可灵、Veo 、Runway 、海螺、万相…，带商品一致性锁，不编造模型参数） - edge-tts 配音、字幕花字、ffmpeg 自动成片 - 平台导出（抖音 / TikTok Shop / 视频号 / 小红书 3:4 / 淘宝主图 1:1 / Amazon 16:9 ）+ 安全区预览 - 直播录像切 9:16 高光、开箱 / 对比评测模板、多语言配音（--lang en ）、封面标题 A/B 安装：npx skills add xianyu110/ecommerce-video-skills，或者 Claude Code 里 /plugin marketplace add xianyu110/ecommerce-video-skills，或者直接把 skills/ 复制到 ~/.claude/skills / ~/.codex/skills / .cursor/skills 。不用 agent 的话，脚本也能单独跑，仓库里带了一个 6 张图的示例。 已知局限： - 静图 Ken Burns 终究不是实拍，适合测款和铺量，跑出数据的品建议补实拍； - 平台的时长 / 大小上限变化太快，没写死，导出时只给提示； - 示例商品是 AI 生成的虚构商品。 欢迎拍砖，尤其是做电商的朋友：哪些平台规格 / 模板最想要？