Wren is in Alpha — free to use while we're building. See planned pricing →
Multi-Model Intelligence

Pick the right brain
for the job.

Most AI coding tools lock you into one model, one speed, one price. ScriptWEAVER supports multiple model tiers so you can choose based on the actual requirements of each task — not just whatever the vendor decided was "good enough."

Speed — fast, low-cost Balanced — everyday workhorse Quality — deep reasoning
The Problem with Single-Model Tools

One model can't be right
for every situation.

A quick two-line bugfix doesn't need a frontier reasoning model costing 10x more and taking 20 seconds. Conversely, a complex architecture decision for your game's economy system probably deserves more deliberate reasoning than a fast, lightweight model can provide.

Most competitors pick one model and call it a day — either because it simplifies their stack or because model pricing is complex. We chose to do the harder work and support multiple tiers because we think developers deserve that control.

Choosing the wrong model isn't just a performance issue — it's wasted time and potentially a wasted budget. I'll recommend a tier, but you always have the override.
W
Wren
Cost Intensity vs. Output Quality
Quality
Cost
Speed
Balanced
Quality

Qualitative illustration. Exact costs to be determined based on usage during final pricing.

Tiers

Three tiers. Different jobs.

Speed
Low latency
Fast

When you need an answer in under two seconds — syntax checks, quick bugfixes, one-liner helpers, rapid iteration on small changes. Speed gets out of your way.

Latency < 2 seconds
Output quality ★★★☆☆
Cost intensity Low
Context window Standard
Best for
Bugfixes Syntax One-liners Fast iteration
Balanced
Speed + depth
Default

The right choice for most everyday tasks. Feature additions, UI work, multi-file refactors, DataStore setups — you get thoughtful output on a reasonable timeline at a moderate cost.

Latency 4–10 seconds
Output quality ★★★★☆
Cost intensity Medium
Context window Large
Best for
Features UI work Refactors Multi-file
Quality
Deep reasoning
Deep

When depth matters more than speed. Architecture decisions, complex multi-system coordination, performance optimization, and security review — tasks that benefit from deliberate, thorough reasoning.

Latency 10–30 seconds
Output quality ★★★★★
Cost intensity High
Context window Very Large
Best for
Architecture Perf Security Complex logic
Side by Side

Full comparison.

Capability Speed Balanced Quality
Latency (approx) Under 2s 4–10s 10–30s
Output quality Good Very Good Excellent
Cost intensity Low Medium High
Context window Standard Large Very Large
Reasoning depth Surface-level Good Deep
Multi-file coordination Limited ✔ Good ✔ Excellent
Architecture & design Not ideal Capable ✔ Best
Bugfixes & quick edits ✔ Best ✔ Good Overkill
UI / feature work Adequate ✔ Best ✔ Good

Qualitative comparison. Specific model versions subject to change as we finalise the product.

Not Sure Which to Pick?

Model Recommender.

Answer three quick questions and I'll suggest the right tier for your task.

Quick Reference

When to use each tier.

⚡ Speed Tier
  • ✔Fixing a nil index error
  • ✔Converting a loop to task.spawn
  • ✔Adding a simple print statement
  • ✔Checking syntax on a new function
  • ✔Writing a utility helper (5–10 lines)
  • ✔Rapid back-and-forth iteration
⚖ Balanced Tier
  • ✔Building a leaderboard system
  • ✔Creating a shop UI
  • ✔Refactoring 2–4 scripts
  • ✔Adding DataStore persistence
  • ✔A round system or game loop
  • ✔Most feature requests
✦ Quality Tier
  • ✔Full economy system design
  • ✔Game-wide performance audit
  • ✔Anti-exploit architecture
  • ✔Large-scale refactor (10+ files)
  • ✔Complex multi-system orchestration
  • ✔Reviewing another AI's output