Claude Opus 4.8
Claude Opus 4.8 is developed by Anthropic. MeetNewAI lists 6 benchmark boards and 3 plans for it.
Plans and real price
Cheapest measured$0.0077per million tokens, measuredClaude Pro · low confidence · 2026-09-14~2026-09-22
API list price$5.00 / $25input / output per million tokens
| Plan | Monthly price | Usable tokens / month | Real price / MTok | Confidence | Data date |
|---|
| Claude Pro | $20 | 2.6B | $0.0077 | low | 2026-09-14~2026-09-22 |
| Claude Max 20x (9/14+) | $200 | 15.7B | $0.013 | low | Aug 1, 2026 |
| Droid Max | $200 | 2B | $0.099 | medium | Sep 29, 2026 |
Data: real-api-pricing · Sources and licenses
Benchmarks
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Opus 4.8 (max) | — | 41.8 | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Code - Opus 4.8 (max) | Claude Code | 23.6 (20.1–27.2) | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Artificial Analysis - Claude Opus 4.8 (max) | Artificial Analysis | 21.7 | — |
| Configuration | Harness | Score | Cost per task |
|---|
| claude-opus-4-8-high | — | 1556.0 (1550.3–1561.7) | — |
| claude-opus-4-8 | — | 1534.0 (1528.3–1539.7) | — |
| Configuration | Harness | Score | Cost per task |
|---|
| Claude Opus 4.8 (High) | — | 6.9 (5.5–8.3) | — |
| Configuration | Harness | Score | Cost per task |
|---|
| mini-swe-agent - claude-opus-4-8 (max) | mini-swe-agent | 59.0 (57.2–60.7) | $13 |
| mini-swe-agent - claude-opus-4-8 (xhigh) | mini-swe-agent | 54.4 (50.6–58.1) | $8.01 |
| mini-swe-agent - claude-opus-4-8 (high) | mini-swe-agent | 51.8 (47.2–56.3) | $4.28 |
| mini-swe-agent - claude-opus-4-8 (medium) | mini-swe-agent | 48.7 (46.4–50.9) | $3.44 |
| mini-swe-agent - claude-opus-4-8 (low) | mini-swe-agent | 40.8 (39.3–42.3) | $2.29 |
Data: real-api-pricing · Sources and licenses
