Archived product note
This page preserves an April 2026 announcement about controls that have since been retired. Webhound now exposes Hound, a research harness built with DeepSeek V4 Pro and GPT-5.4 across planning, execution, verification, and assembly. Hound is not a selectable model or mode and does not resolve to a single provider backend. You choose the question and dollar budget; the budget controls research effort. See the current setup guide.
We're raising the minimum budget for Pro sessions. Pro reports now start at $10 instead of $3, and Pro datasets at $5. Flash reports drop to $2 so quick scans stay accessible. Flash datasets stay at $1. Max mode is roughly 2–3× the normal minimum.
| Session | Model | Standard | Max mode |
|---|---|---|---|
| Report | Flash | $2 | $8 |
| Report | Pro | $10 | $25 |
| Dataset | Flash | $1 | $3 |
| Dataset | Pro | $5 | $15 |
Changed values shown in red. Everything else is the same as before.
The reason is simple. Pro uses a much smarter model, and at $3 it wasn't enough budget to finish a real report. People were starting Pro sessions expecting Pro-quality output and running out of money before the agent got anywhere. The new floors make sure every Pro session actually produces something worth reading.
Flash is cheaper to run, so the floor for a Flash report comes down to $2 — still enough to do a real scan, just lower friction to start.
Which one should you use
Pro is for reports. It's more accurate and consistent — what you want when the output has to hold up. Flash is for datasets.It's fast and cheap, which is what you want when you need lots of rows and occasional errors are easy to spot.
Flash can make mistakes that Pro usually doesn't. If you're writing a memo, strategy doc, or anything someone will push back on, use Pro. If you're building a list, directory, or catalog, use Flash.
We also rebuilt the pricing page to show what each budget actually buys — $10 vs $25 vs $100, in minutes and sources, based on real sessions from the last month. Questions: team@webhound.ai.