|
92 | 92 | <nav id="primary-nav" aria-label="Primary"> |
93 | 93 | <a href="#why">Why</a> |
94 | 94 | <a href="#what">What</a> |
| 95 | + <a href="#github-models">Compare</a> |
95 | 96 | <a href="#quickstart">How</a> |
96 | 97 | <a href="#features">Features</a> |
97 | 98 | <a href="#faq">FAQ</a> |
@@ -235,6 +236,118 @@ <h3> |
235 | 236 | </article> |
236 | 237 | </section> |
237 | 238 |
|
| 239 | + <section |
| 240 | + id="github-models" |
| 241 | + class="container section section-github-models" |
| 242 | + aria-labelledby="github-models-title" |
| 243 | + > |
| 244 | + <p class="eyebrow">How it compares</p> |
| 245 | + <h2 id="github-models-title"> |
| 246 | + From prompt experiments to production prompt assets |
| 247 | + </h2> |
| 248 | + <p class="section-intro"> |
| 249 | + GitHub Models is great for exploring prompts, comparing models, and |
| 250 | + running evaluations inside GitHub. PromptOpsKit focuses on what |
| 251 | + happens when that behavior needs to ship inside your application. |
| 252 | + </p> |
| 253 | + <p class="section-intro section-intro-tight"> |
| 254 | + <strong> |
| 255 | + GitHub Models helps you experiment with prompts in GitHub. |
| 256 | + PromptOpsKit helps you ship prompt behavior inside your application. |
| 257 | + </strong> |
| 258 | + </p> |
| 259 | + |
| 260 | + <div class="grid two comparison-grid"> |
| 261 | + <article class="card comparison-card"> |
| 262 | + <h3>GitHub Models</h3> |
| 263 | + <ul> |
| 264 | + <li>Prompt editor and playground inside GitHub</li> |
| 265 | + <li>Model comparison and structured evaluations</li> |
| 266 | + <li> |
| 267 | + Repository-stored <code>.prompt.yml</code> or |
| 268 | + <code>.prompt.yaml</code> files |
| 269 | + </li> |
| 270 | + <li> |
| 271 | + Great for experimentation, collaboration, evals, and model |
| 272 | + selection |
| 273 | + </li> |
| 274 | + </ul> |
| 275 | + <p class="comparison-note"> |
| 276 | + GitHub's docs describe prompt files, model parameters, test data, |
| 277 | + evaluators, and CLI evals with <code>gh models eval</code>. |
| 278 | + </p> |
| 279 | + </article> |
| 280 | + |
| 281 | + <article class="card comparison-card comparison-card-strong"> |
| 282 | + <h3>PromptOpsKit</h3> |
| 283 | + <ul> |
| 284 | + <li>Markdown prompt assets designed for application runtime</li> |
| 285 | + <li> |
| 286 | + Input hardening, environment overrides, composition, and |
| 287 | + validation |
| 288 | + </li> |
| 289 | + <li> |
| 290 | + Provider-specific request rendering for your own app code |
| 291 | + </li> |
| 292 | + <li> |
| 293 | + Great for shipping prompt behavior through Git, CI, and |
| 294 | + production deploys |
| 295 | + </li> |
| 296 | + </ul> |
| 297 | + <p class="comparison-note"> |
| 298 | + It renders provider-shaped request bodies while your app keeps |
| 299 | + control of SDKs, auth, retries, routing, observability, and |
| 300 | + billing. |
| 301 | + </p> |
| 302 | + </article> |
| 303 | + </div> |
| 304 | + |
| 305 | + <div class="github-models-points"> |
| 306 | + <article> |
| 307 | + <h3>Runtime-first</h3> |
| 308 | + <p> |
| 309 | + Turn prompt files into provider-ready request bodies your |
| 310 | + application can execute with your own deployment model. |
| 311 | + </p> |
| 312 | + </article> |
| 313 | + <article> |
| 314 | + <h3>Markdown-native</h3> |
| 315 | + <p> |
| 316 | + Keep long system instructions, templates, notes, and examples |
| 317 | + readable in Markdown, with YAML front matter for metadata. |
| 318 | + </p> |
| 319 | + </article> |
| 320 | + <article> |
| 321 | + <h3>Input hardening</h3> |
| 322 | + <p> |
| 323 | + Define required variables, size limits, trimming rules, and |
| 324 | + allow/deny regex checks next to the prompt. |
| 325 | + </p> |
| 326 | + </article> |
| 327 | + <article> |
| 328 | + <h3>Bring your own transport</h3> |
| 329 | + <p> |
| 330 | + Render OpenAI, Anthropic, Gemini, OpenRouter, or OpenAI Responses |
| 331 | + request bodies, then send them through your existing stack. |
| 332 | + </p> |
| 333 | + </article> |
| 334 | + <article> |
| 335 | + <h3>Environment-aware</h3> |
| 336 | + <p> |
| 337 | + Keep dev, staging, production, and tier-specific model settings |
| 338 | + close to the prompt and reviewable in Git. |
| 339 | + </p> |
| 340 | + </article> |
| 341 | + <article> |
| 342 | + <h3>Composable</h3> |
| 343 | + <p> |
| 344 | + Reuse shared instructions, policies, and template fragments |
| 345 | + across prompts without copying the same blocks everywhere. |
| 346 | + </p> |
| 347 | + </article> |
| 348 | + </div> |
| 349 | + </section> |
| 350 | + |
238 | 351 | <section id="demo" class="container section section-demo"> |
239 | 352 | <div class="demo-copy"> |
240 | 353 | <p class="eyebrow">See it working</p> |
@@ -624,6 +737,22 @@ <h3>Usage-based billing and entitlement checks</h3> |
624 | 737 | <section id="faq" class="container section"> |
625 | 738 | <h2>Frequently asked questions</h2> |
626 | 739 | <div class="grid two"> |
| 740 | + <article class="card"> |
| 741 | + <h3>How is this different from GitHub Models?</h3> |
| 742 | + <p> |
| 743 | + GitHub Models is great for experimenting with prompts, comparing |
| 744 | + models, and running evaluations inside GitHub. PromptOpsKit |
| 745 | + focuses on the application runtime layer: Markdown-based assets |
| 746 | + with structured metadata, input validation, environment overrides, |
| 747 | + composition, and provider-specific request rendering. |
| 748 | + </p> |
| 749 | + <p> |
| 750 | + In practice, GitHub Models helps you decide what prompt and model |
| 751 | + behavior you want. PromptOpsKit helps you ship that behavior |
| 752 | + inside your app without replacing your SDK, gateway, auth, |
| 753 | + retries, observability, or deployment pipeline. |
| 754 | + </p> |
| 755 | + </article> |
627 | 756 | <article class="card"> |
628 | 757 | <h3>What happens when I need to change models?</h3> |
629 | 758 | <p> |
|
0 commit comments