Plans and real price
Cheapest measured$0.0055per million tokens, measuredClaude Max 5x (9/14+) · medium confidence · 2026-09-29~2026-10-03
API list price$4.00 / $20input / output per million tokens
| Plan | Monthly price | Usable tokens / month | Real price / MTok | Confidence | Data date |
|---|
| Claude Max 5x (9/14+) | $100 | 18.3B | $0.0055 | medium | 2026-09-29~2026-10-03 |
| Claude Max 20x (9/14+) | $200 | 31.5B | $0.0063 | medium | 2026-09-22~2026-09-27 |
| Claude Pro | $20 | 3.1B | $0.0065 | high | 2026-09-22~2026-09-27 |
| Claude Team Premium | $125 | 12.6B | $0.0099 | low | 2026-09-30~2026-10-06 |
| Devin Max | $200 | 6.7B | $0.030 | medium | Sep 23, 2026 |
| Droid Max | $200 | 2.5B | $0.079 | medium | Sep 29, 2026 |
| Droid Pro | $20 | 215M | $0.093 | low | Sep 28, 2026 |
| Cursor Ultra (Other Models) | $200 | 896M | $0.223 | medium | 2026-09-14~2026-10-05 |
| Claude Opus 5.5 API | — | — | $0.419 | high | Sep 23, 2026 |
Data: real-api-pricing · Sources and licenses
API price
| Input / 1M | $4.00 |
|---|
| Cached / 1M | $0.200 |
|---|
| Output / 1M | $20 |
|---|
At 1,000 requests a day (2,000 input and 500 output tokens each), about $540 a month.
API cost calculator · Anthropic API pricing
Data: real-api-pricing · Sources and licenses
Benchmarks
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Opus 5.5 (max with fallback) | — | 57.6 | — |
| Claude Opus 5.5 (xhigh with fallback) | — | 56.0 | — |
| Claude Opus 5.5 (high with fallback) | — | 53.6 | — |
| Claude Opus 5.5 (medium with fallback) | — | 51.2 | — |
| Claude Opus 5.5 (low with fallback) | — | 42.3 | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Code - Claude Opus 5.5 (default) - self-reported [vendor self-report] · self-reported | Claude Code | 66.4 | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Artificial Analysis - Claude Opus 5.5 (max with fallback) | Artificial Analysis | 59.6 | — |
| Artificial Analysis - Claude Opus 5.5 (xhigh with fallback) | Artificial Analysis | 59.6 | — |
| Artificial Analysis - Claude Opus 5.5 (high with fallback) | Artificial Analysis | 56.6 | — |
| Artificial Analysis - Claude Opus 5.5 (medium with fallback) | Artificial Analysis | 52.5 | — |
| Artificial Analysis - Claude Opus 5.5 (low with fallback) | Artificial Analysis | 31.3 | — |
| Configuration | Harness | Score | Cost per task |
|---|
| claude-opus-5.5-max | — | 1818.0 (1801.5–1834.5) | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Opus 5.5 (High) | — | 13.8 (11.5–16.0) | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Code - Opus 5.5 (max) | Claude Code | 66.0 | $13 |
Data: real-api-pricing · Sources and licenses
Codex Radar runs
| Effort | Harness | Radar IQ | Avg cost (USD) | Avg minutes |
|---|
| low | codex | 150.0 | $0.554 | 3.5 |
| medium | codex | 150.0 | $0.665 | 3.3 |
| xhigh | codex | 150.0 | $2.61 | 11.0 |
| max | codex | 150.0 | $22 | 84.3 |
Data: codexradar · Sources and licenses
