Anthropic
Claude Opus 5
- Frontier list
- $5.00 / $25.00
- You · in / out
- $6.50 / $32.50
Model rate card
What the models cost. What you pay.
Frontier LLM list price, plus 30% to Bench.
You buy Fuel as a monthly tank. These are the list prices behind Fuel.
Flagships · list vs you
Input and output rates per million tokens. “You” is the public list rate with Bench's fixed addition applied. Grok bills the whole request at a higher rate from 200,000 input tokens. OpenAI 5.6 does the same from 272,000. Anthropic does not. The calculator uses those steps.
OpenAIPromo at least through November 21, 2026
GPT-5.6 Sol
- Frontier list
- $4.00 / $20.00
- You · in / out
- $5.20 / $26.00
Promo at least through November 21, 2026. Standard list $5.00 / $30.00 → you $6.50 / $39.00.
xAI
Grok 4.6
- Frontier list
- $2.00 / $6.00
- You · in / out
- $2.60 / $7.80
Same job, both bills
Pick a token shape to compare the model provider's list bill with this rate card.
- Input tokens
- 40,000
- Output tokens
- 6,000
- Cached share
- 20%
- Plus to Bench
- 30%
| Model | Frontier list | Plus to Bench | You pay |
|---|---|---|---|
| GPT-5.6 Luna | $0.0138 | $0.0041 | $0.0179 |
| Grok 4.3 | $0.06 | $0.02 | $0.07 |
| Claude Haiku 4.5 | $0.06 | $0.02 | $0.08 |
| Grok 4.5 | $0.10 | $0.03 | $0.13 |
| Grok 4.6 | $0.10 | $0.03 | $0.14 |
| Claude Sonnet 5 | $0.13 | $0.04 | $0.16 |
| GPT-5.6 Terra | $0.14 | $0.04 | $0.18 |
| GPT-5.6 Sol | $0.25 | $0.08 | $0.33 |
| Claude Opus 5 | $0.31 | $0.09 | $0.41 |
| Fable 5 | $0.63 | $0.19 | $0.82 |
The seats we run
| Seat | Model | List · in / out | You · in / out |
|---|---|---|---|
| Lead (unclear / T4) | Fable 5 | $10.00 / $50.00 | $13.00 / $65.00 |
| Lead (known shape) / panel | Claude Opus 5 | $5.00 / $25.00 | $6.50 / $32.50 |
| Pre-map / workers | Claude Sonnet 5 | $2.00 / $10.00 | $2.60 / $13.00 |
| Trivial | Claude Haiku 4.5 | $1.00 / $5.00 | $1.30 / $6.50 |
| Builder + first adversarial / Daybreak Blue | GPT-5.6 Sol | $4.00 / $20.00 | $5.20 / $26.00 |
| Parent | Grok 4.6 | $2.00 / $6.00 | $2.60 / $7.80 |
Full current lineup
| From | Model | List · in / out | You · in / out |
|---|---|---|---|
| Anthropic | Fable 5 | $10.00 / $50.00 | $13.00 / $65.00 |
| Anthropic | Claude Opus 5 | $5.00 / $25.00 | $6.50 / $32.50 |
| Anthropic | Claude Sonnet 5 | $2.00 / $10.00 | $2.60 / $13.00 |
| Anthropic | Claude Haiku 4.5 | $1.00 / $5.00 | $1.30 / $6.50 |
| OpenAI | GPT-5.6 Sol | $4.00 / $20.00 | $5.20 / $26.00 |
| OpenAI | GPT-5.6 Terra | $2.00 / $12.00 | $2.60 / $15.60 |
| OpenAI | GPT-5.6 Luna | $0.20 / $1.20 | $0.26 / $1.56 |
| xAI | Grok 4.6 | $2.00 / $6.00 | $2.60 / $7.80 |
| xAI | Grok 4.5 | $2.00 / $6.00 | $2.60 / $7.80 |
| xAI | Grok 4.3 | $1.25 / $2.50 | $1.63 / $3.25 |
Grok 4.1 Fast is available on Fuel, but has no public model list price to publish here. See how the lineup runs on Fuel.
Why 30% to Bench?
We pay xAI, Anthropic, and OpenAI first. The job can finish before your card settles.
If a payment fails after the tokens are burned, we take that risk.
The 30%covers that risk and the routing so you don't pick models. It isn't 30%profit. What's left after the stack is a thin keep.