100% renewable energy inference, three ways.
TreadLightlyAI is a thinking tool with 7 modes. Every prompt has a cost, we show you exactly what you use. Pick the plan that fits.
Free
Try the thinking tool, no card required.
- Green Tiny + Green Small models
- 1M tokens/month combined
- Analyze image and PDF files
- 200 MB file storage
- 2 projects
- Unlimited diagrams
- Export conversations
Sign up to create your free account
Starter
Start with a 7-day free trial
Add Green Large for stronger reasoning.
- Everything in Free
- Add Green Large model
- 5M tokens/month combined
- 1 GB file storage
- 5 projects
Builder
Start with a 7-day free trial
The full-featured green tier.
- Everything in Starter
- Unlimited Green Tiny
- 8M Green Small + 4M Green Large tokens
- 4 GB file storage
- 20 projects
Advanced
Start with a 7-day free trial
Adds Claude Haiku + Sonnet for safety-leadership capability.
- Everything in Builder
- Unlimited Green models
- 2M Claude tokens/month (1M Haiku + 1M Sonnet)
- 20 GB file storage
- 20 projects
Why Advanced exists
Free, Starter, and Builder run on 100% renewable energy inference. Advanced adds Anthropic Claude (Haiku and Sonnet) for work that needs stronger reasoning, long-context analysis, or Anthropic's safety tuning. Claude runs on infrastructure we can't certify as renewable, so Advanced is where that compromise lives. Unlimited Green models, 2M Claude tokens per month (1M Haiku + 1M Sonnet), same per-prompt cost transparency.
See the AI Transparency Card for the other compromises we make.
Smart Token Management
Context Size Controls
Control your context window size, trim older messages, or start fresh, so you spend tokens on what actually matters.
Real-Time Usage Tracking
Your token usage is displayed in real time, so there are no surprises.
What's Included in Every Plan
7 Thinking Modes
Each mode changes how TreadLightlyAI works with you, from skeptical stress-tester to creative collaborator to forensic auditor. All seven, on every plan.
| Mode | Stance | Best For |
|---|---|---|
| Challenge | Stress-test and find weaknesses | Validating designs, catching blind spots |
| Research | Investigate with epistemic transparency | Research, competitive analysis |
| Write | Produce content ready to use | Drafting, editing, content creation |
| Teach | Scaffold understanding with examples | Learning new topics, building knowledge |
| Plan | Structure and sequence work | Project planning, defining next steps |
| Audit | Evaluate real-world viability | Cost analysis, feasibility checks |
| Balanced | Direct, helpful generalist | General assistance, open-ended tasks |
Frequently Asked Questions
How does the Free plan work?
The Free plan includes Green Tiny and Green Small with 1M tokens per month combined. No credit card required.
Can I change plans anytime?
Yes. Upgrade or downgrade from your profile page.
What happens if I exceed my token limit?
You can see your current usage on your profile page. Once you hit your limit, you won't be able to send new messages until you upgrade or your limit resets at the start of the next billing period. No hidden charges.
Which model should I use?
- Green Small: Fast, good for everyday tasks. The default.
- Green Large: Stronger eco reasoning. Still 100% green.
- Haiku: Fast Claude model when you need Anthropic quality. Advanced only.
- Sonnet: Claude's strongest reasoning for hard problems. Advanced only.
What are the exact token limits?
Each model has its own monthly allocation:
| Plan | Green Tiny | Green Small | Green Large | Haiku | Sonnet |
|---|---|---|---|---|---|
| Free | 500K | 500K | — | — | — |
| Starter | 2M | 2M | 1M | — | — |
| Builder | ∞ | 8M | 4M | — | — |
| Advanced | ∞ | ∞ | ∞ | 1M | 1M |
Unlimited means no monthly cap. Service stability requires per-minute limits on every tier.
Do you offer refunds?
If you're not satisfied, contact us at support@treadlightly.ai. We handle refund requests case by case. For subscription renewals, refund requests must be made within 48 hours of the charge.
Is my data secure?
We never sell your data, never share it for marketing, and never use your conversations to train AI models. Your conversations are yours. See our Privacy Policy for the full picture.
How does energy reporting work?
Most conversations track their energy use, visible from the conversation menu.
- Measured: GreenPT reports energy per call, streaming and non-streaming alike, in milliwatt-hours.
- Estimated: Claude and some green models use a token-based estimate, marked with a tilde (~).
- Pre-send: A relative size estimate (S/M/L/XL) before you send.
More in help.