Tiers & Pricing
Choose performance, not models. TierUp abstracts model selection into 4 simple tiers. Pick the tier that matches your task — we handle routing to the best underlying model automatically.
We optimize model selection per tier — you always get the best performance-per-dollar. All tiers priced ~50% below retail.
Pricing
| Tier | Label | Quality Level | Input / 1M | Output / 1M | vs Model | Retail | Savings |
|---|---|---|---|---|---|---|---|
| 1 | Speed | Fast, simple tasks (Haiku or similar) | Free | Free | Haiku 4.5 | $1.00 / $5.00 | 100% |
| 2 | Balance | Best all-rounder for chat & code (Sonnet or similar) | $1.50 | $7.50 | Sonnet 4.6 | $3.00 / $15.00 | 50% |
| 3 | Intelligence | Highest general intelligence (Opus or similar) | $2.50 | $12.50 | Opus 4.6 | $5.00 / $25.00 | 50% |
| 4 | Reasoning | Deep chain-of-thought reasoning (o3 or similar) | $1.00 | $4.00 | OpenAI o3 | $2.00 / $8.00 | 50% |
Prices are per 1 million tokens.
How Tiers Work
Tiers represent different types of intelligence, not ascending quality levels. You pass "tier": 2 in your request. We route to the best available model for that tier — you never see which underlying model ran. The response always shows a stable TierUp model name such as "model": "tierup-balance".
As frontier models improve, we update our routing automatically. Your code stays the same; your results only get better.
Choosing a Tier
| Tier | Best for |
|---|---|
| 1 — Speed | Classification, extraction, formatting, quick Q&A, simple tasks |
| 2 — Balance | Production chatbots, content generation, coding, analysis (best all-rounder) |
| 3 — Intelligence | Complex coding, nuanced analysis, research-grade work, long-form writing |
| 4 — Reasoning | Math problems, logic puzzles, scientific reasoning, multi-step proofs |
Cost Calculation
Each request is billed based on the number of input and output tokens consumed:
+ (output_tokens / 1,000,000) × output_price
The result is rounded to the nearest cent. You can view a per-request breakdown in the Usage dashboard.
