Lowest price on the platform, built for always-on pipelines.
Overview
HacxGPT Lightning Flash is the entry point to the platform: a small, fast model tuned for latency and cost. It handles classification, extraction, tool routing, and bulk text operations at a rate that makes always-on pipelines practical - and it answers security work without lectures in the same breath.
It accepts the two light reasoning levels (medium, high) for tasks that need a moment of deliberation. It is the wrong tool for long-horizon research or deep multi-step work; reach for Lightning or Ninja Red when depth matters more than throughput.
HacxGPT models are in beta: production in use, experimental, evolving. We publish the basics - benchmarks, context window, pricing, and how the models are made - and add detail as each model matures. If you need specifics we haven't published yet, ask.
Capabilities
Refusal behavior is not a sticker on top of a model; it lives in the model's weights. HacxGPT models are built by locating that behavior and reducing it in the weights, then fine-tuned for instruction-following so fewer refusals never means worse answers. The result holds across sessions instead of depending on prompt tricks.
Make the call
curl https://api.hacxgpt.com/v1/chat/completions \
-H "Authorization: Bearer $HACX_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "lightning-flash",
"reasoning_effort": "medium",
"messages": [{"role": "user", "content": "..."}]
}'Flash is evaluated on throughput and cost-per-million rather than exam suites; quality benchmarks publish as the beta progresses.
Model questions
Long-horizon research, deep reasoning chains, or heavy agentic work. Use Lightning or Ninja Red when depth matters more than throughput.
Yes. It accepts the two light reasoning levels, medium and high, set through reasoning_effort. Deeper thinking belongs on bigger models.
Get started
One API key reaches every model on the platform.