Model profile / Anthropic
Anthropic's documented model for complex agentic coding and enterprise work.
Four areas of evidence
Mathematical and scientific problem solving.
Score uses the admitted direct native evidence under the selected method.
Software engineering and programming.
Score uses the admitted direct native evidence under the selected method.
Factual reliability and instruction following.
Score uses the admitted direct native evidence under the selected method.
Tool use, workflows and professional tasks.
Score uses the admitted direct native evidence under the selected method.
Current index snapshot · 2026-09-16. Each category uses its own scale; scores across categories are not directly comparable. Evidence counts describe the retained Capabilities inputs; a family can contain several tasks or configurations. A point lead does not establish superiority.
Performance in context
Claude Opus 4.8 is highlighted wherever a compatible measurement is available.
Lumina Capabilities Index
The current Capabilities cohort, with this model highlighted.
Chart loads as you explore
Lumina Capabilities Index
The current Capabilities cohort, with this model highlighted.
Chart loads as you explore
Explore the detail
Score uses the admitted direct native evidence under the selected method.
Score uses the admitted direct native evidence under the selected method.
Score uses the admitted direct native evidence under the selected method.
Score uses the admitted direct native evidence under the selected method.
Score uses the admitted direct native evidence under the selected method.
| Benchmark | Result | Published configuration | Source |
|---|---|---|---|
| AA Long Context Reasoningstandard | 73% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis AA-LCR evaluation | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| CritPtstandard | 20.857% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis CritPt evaluation | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| GDPval-AA v2v2 | 53.894% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis GDPval-AA v2 using Stirrup | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| GPQA Diamonddiamond | 92.02% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis GPQA Diamond evaluation | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| Humanity's Last Examtext-only current | 48.656% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis text-only Humanity's Last Exam evaluation | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| SciCoderolling | 53.472% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis SciCode evaluation | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · reference-only |
| τ³-Banking3 | 34.227% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis tau3-Banking evaluation | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| Terminal-Bench2.1 | 84.644% | Claude Opus 4.8 (max; Artificial Analysis independent run)Artificial Analysis Terminal-Bench v2.1 in e2b | Claude Opus 4.8 individual evaluations ↗Observed 2026-08-27 · checked 2026-08-27independently-verified · ranking-eligible |
| Artificial Analysis Agentic Index2026 | 47.18 index | Exact BenchLM registry variant Claude Opus 4.8; bulk export does not retain a complete upstream harness configuration.Source-native system | BenchLM public datasets — 21 July 2026 ↗Observed 2026-07-21 · checked 2026-07-21source-checked · reference-only |
| Artificial Analysis Agentic Index2026 | 47.18 index | Exact BenchLM registry variant Claude Opus 4.8; bulk export does not retain a complete upstream harness configuration.Source-native system | BenchLM public datasets — 2026-07-27 ↗Observed 2026-07-27 · checked 2026-07-27source-checked · reference-only |
Specialist evidence
Source-native operational evidence for Claude Opus 4.8. Exact configurations remain separate. Costs are comparable only within the same selected benchmark/evaluation, and these rows never enter Overall Score.
| Evaluation | Exact configuration | Performance | Cost / task | Tokens / task | Execution |
|---|---|---|---|---|---|
| Artificial Analysis Coding Agents · 1.4 · composite Artificial Analysis Coding Agents · checked 2026-08-29 | claude-opus-4-8-epoch-claude-opus-4-8-low Claude Code - Opus 4.8 (low) | 49.0 | $2.18 | 5214050 | 68.2 steps |
| Artificial Analysis Coding Agents · 1.4 · composite Artificial Analysis Coding Agents · checked 2026-08-29 | claude-opus-4-8-high Claude Code - Opus 4.8 (high) | 57.6 | $3.78 | 9188691 | 104.8 steps |
| Artificial Analysis Coding Agents · 1.4 · composite Artificial Analysis Coding Agents · checked 2026-08-29 | claude-opus-4-8-max Claude Code - Opus 4.8 (max) | 62.1 | $7.72 | 17920579 | 165 steps |
| CursorBench · 3.2 · main CursorBench · checked 2026-08-29 | claude-opus-4-8-epoch-claude-opus-4-8-low Opus 4.8 Low | 53.1% | $2.02 | 19624 | 27 steps |
| CursorBench · 3.2 · main CursorBench · checked 2026-08-29 | claude-opus-4-8-high Opus 4.8 High | 58.0% | $3.15 | 33548 | 33 steps |
| CursorBench · 3.2 · main CursorBench · checked 2026-08-29 | claude-opus-4-8-max Opus 4.8 Max | 62.3% | $5.77 | 71411 | 44 steps |
| Terminal-Bench 2.1 · 2.1 · verified Terminal-Bench 2.1 · checked 2026-08-29 | claude-opus-4-8-high Claude Code; reasoning high; Terminal-Bench 2.1 verified submission. | 78.9% | $3.22 | — | — |
