Cost per 1M token-equivalent $1.75
- CP value
- No data
- Equivalent cost / 1M
- $1.75
- Model allowance (theoretical)
- Pay as you go
Same workload comparison
Start with what this workload buys you
The same token mix is applied to each plan’s prices and model allowance. Check shared pools and time windows on the plan page.
Comparison example: 12,000 tokens/request · 83% cache read
Example workload (12,000 tokens/request)
- Input
- 1,000
- Cache read
- 10,000
- Output
- 1,000
- Cache write
- 0
02Prices and allowances
Keep prices, requests and plan windows in one evidence ledger
Expand the ledger to verify four token prices, cost per request, model allowance and theoretical requests for this workload.
Open prices, capacity and plan windows
| Metric | OpenCode Zen Claude Sonnet 4.6 |
|---|---|
| Input / 1M | $3.00 |
| Cache read / 1M | $0.30 |
| Output / 1M | $15.00 |
| Cache write / 1M | $3.75 |
| Cost / request Recalculated | $0.02100012,000 tokens/request |
| Model allowance Published | Pay as you goNo plan allowance; billed per request |
| Requests / month Recalculated | Pay as you goNo plan allowance; compare its rates instead |
| 5-hour limit Published | —Pay as you go |
| Weekly limit Published | —Pay as you go |
| Monthly limit Published | —Pay as you go |
03Official reference rows
The official reference behind each plan
Compare each provider’s reference token mix and published estimates below.
OpenCode Zen View the provider’s reference workload
| Reference input / request | 1,000 |
|---|---|
| Reference cache read / request | 10,000 |
| Reference output / request | 1,000 |
| Reference cache write / request | 0 |
| Reference volume / request | 12,000 |
| Cost per request | $0.021000 |
| Model monthly allowance | Pay as you go |
| Capacity basis | No published reference profile; the site common workload is used instead |
Official source: OpenCode Zen · Snapshot captured: 2026-09-06
04Artificial Analysis external scores
External benchmarks with their source version
Benchmarks are from Artificial Analysis. See the source version and matched model below.
intelligence36.8Intelligence rank: 135
codingNo dataCoding rank: No data
agenticNo dataAgentic rank: No data
Scores describe the model’s performance on the named benchmarks. Results on your own tasks also depend on tools, prompts and execution environment.