dense · 0.8B parameters
Qwen3-0.6B
chat3 recorded quants
Compare recorded files, memory headroom, and estimated decode speed.
Find compatible hardware →Independent local AI hardware guide
Big ideas.
Your computer.
Will it run?
Find the right hardware for the AI you want to run at home. Check memory, compare models, and see the math behind every estimate.
✓ Official source records✓ No cloud inference✓ Clear assumptions
Will it run?
Memory + speed estimateLoading the local dataset…
estimated tokens / sec
Model weightsKV cacheOverhead
Estimated, not benchmarked. Calculations stay in this browser. See the methodology.
150Open-weight model records
44Hardware configurations
1721Compatible pair pages
ZeroCalculator API calls
Start with a model.
All models →dense · 4.0B parameters
Qwen3-4B
chat3 recorded quants
Compare recorded files, memory headroom, and estimated decode speed.
Find compatible hardware →dense · 14.8B parameters
Qwen3-14B
chat3 recorded quants
Compare recorded files, memory headroom, and estimated decode speed.
Find compatible hardware →A clearer kind of estimate
Memory first.
Then the bigger picture.
Model weights are only part of the equation. We account for context cache and runtime overhead, then compare the result with a usable memory budget.
Read the full methodology →- Use the actual model files, including every recorded shard.
- Reserve room for your system and the context you need.
- Label speed as estimated, with the formula and its limits.
A little guidance goes a long way.
All guides →Buying guide
A private AI computer under $1,000
A sourced shortlist that starts with memory, uses complete-system prices, and makes room for uncertainty.
Read the guide →Buying guide
A private AI computer under $2,000
A sourced shortlist that starts with memory, uses complete-system prices, and makes room for uncertainty.
Read the guide →Buying guide
A private AI computer under $4,000
A sourced shortlist that starts with memory, uses complete-system prices, and makes room for uncertainty.
Read the guide →