AISA Turbo: parallel speed and auto-tuning on rate limits

What it is

AISA Turbo speeds up large 1-Click SEO and Bulk runs by sending several AI requests in parallel instead of one at a time. You pick a multiplier (×2, ×4, ×8, ×10, ×20) and AISA fires that many generations concurrently, so optimising thousands of pages can take minutes instead of hours. Turbo also auto-tunes: when a provider pushes back with a rate-limit/overload response (HTTP 429, 503 or 529), AISA backs the speed down automatically (an AIMD-style additive-increase/multiplicative-decrease loop) so the batch keeps running instead of failing.

Where to find it

AISA → Settings, under the analysis settings: the AISA Turbo (parallel generation) dropdown with options Off (1 at a time), ×2, ×4 (recommended), ×8, ×10, ×20. The effect is visible during 1-Click SEO and Bulk SEO runs, where the progress bar advances faster at higher multipliers.

Requirements & tier

Available with both BYOK providers and Cloud, and is most useful on large batches (1-Click / Bulk).
A provider that tolerates parallel load. DeepSeek is best for big batches (generous limits, no hard daily cap); OpenAI has a per-tier daily request cap that high Turbo can exhaust.
Combine with the multi-key pool for more real headroom: multiple independent keys = more parallel requests before hitting limits.

How to use

Go to AISA → Settings and set AISA Turbo to a multiplier. ×4 is the recommended starting point.
For very large sites, use DeepSeek for the text batch (its limits suit high speed) and save OpenAI/Gemini for small volumes or images.
Optionally add extra keys in the multi-key pool so Turbo can spread requests across them.
Run 1-Click SEO or Bulk SEO. Watch the live progress, scores and credits/tokens update in real time.
If you see slowdowns, let Turbo auto-tune — it will reduce concurrency on 429/503/529 responses on its own. You can also lower the multiplier manually.
Stop any time; finished items stay saved and you can resume with "complete the remaining".

Questions people ask

What multiplier should I use? ×4 is the recommended default. Go higher only on DeepSeek or when you have several independent keys; pull back if your provider rate-limits.
What happens on a 429? AISA marks the call as rate-limited and Turbo automatically decreases concurrency (AIMD), so the run continues at a safe pace instead of erroring out.
Does Turbo cost more? No — it changes only how fast requests go out, not how many tokens/credits each one uses. Total cost is the same; it just finishes sooner.
Why does OpenAI choke at high Turbo? Its per-tier daily request cap (Tier 1 ~10,000/day) can run out under heavy parallel load. Use DeepSeek for huge batches or raise your OpenAI tier.
Does adding more keys really help? Independent keys help a lot. But several keys from the same account often share one limit, depending on the provider tier.
Can I run Turbo in the background? 1-Click has a "continue in background" hand-off so the batch keeps going with the page closed.

Troubleshooting

If a batch keeps getting rate-limited even after auto-tuning, lower the Turbo multiplier, switch the batch to DeepSeek, and add independent keys to the pool. When a key hits a 429, AISA puts that specific key into a short cooldown and prefers the others, so a pool genuinely smooths things out. If OpenAI blocks until reset, it is the daily request cap — that is a provider limit, not an AISA error. If speed feels capped despite a high multiplier, you are probably bound by a single key's per-minute limit; the fix is more keys, not a higher number.

Note on speed: Processing time depends on two things: your AI provider's response speed and the server where the plugin runs (PHP workers, memory and hosting). A slow provider or a constrained host makes large batches take longer — use AISA Turbo and run in the background to mitigate it.

Multi-key pool: spread load across several keys
Supported AI providers and which one to choose
Choosing the model and understanding costs
Connect an AI provider and add your API key