Loading benchmark catalogue
Compare performance benchmarks, coding intelligence, reasoning evaluations, context limits, and API economics between Claude Opus 5.5 (Anthropic) and Claude Haiku 5.5 (Anthropic).
| Metric / Capability | Claude Opus 5.5 | Claude Haiku 5.5 |
|---|---|---|
| Developer / Organization | Anthropic | Anthropic |
| Release Date | 2026-09-22 | 2026-10-07 |
| Deployment / Weights | Proprietary Cloud API | Proprietary Cloud API |
| Context Window | 1M tokens (1,000,000 tokens) | 1M tokens (1,000,000 tokens) |
| Input Pricing (/1M Tokens) | $4 | $0.1 |
| Output Pricing (/1M Tokens) | $20 | $0.5 |
| Benchmark / Domain | Claude Opus 5.5 | Claude Haiku 5.5 | Advantage |
|---|---|---|---|
Artificial Analysis Intelligence Index Composite index aggregating 10+ evaluations of general reasoning and knowledge. | 57.6% | 43.4% | Claude Opus 5.5 (+33%) |
Artificial Analysis Coding Agent Index v2.0 Composite index of coding-agent systems (model + harness) across FrontierCode v1.1, SWE-Atlas-QnA, and Terminal-Bench v4. | 61.7% | 37.3% | Claude Opus 5.5 (+65%) |
SWE-bench Pro Harder, contamination-resistant successor to SWE-bench spanning multiple languages. | 89.9% | — | — |
Terminal-Bench 4.0 Sandboxed command-line tasks requiring shell and file-system operation. | 66.4% | 39.2% | Claude Opus 5.5 (+69%) |
ARC-AGI-2 Novel visual grid puzzles requiring abstract rule induction. | 91.7% | 61.3% | Claude Opus 5.5 (+50%) |
Humanity's Last Exam Expert-written questions across hundreds of disciplines designed to resist saturation. | 64.4% | 45.9% | Claude Opus 5.5 (+40%) |
FrontierMath v2 (Tiers 1-3) Unpublished, research-level mathematics problems across Tiers 1-3. | 91.2% | 75.1% | Claude Opus 5.5 (+21%) |
Design Arena ELO Human-preference ELO across UI, 3D, data-visualization, and game-dev generations. | 1468 | 1344 | Claude Opus 5.5 (+9%) |
Claude Opus 5.5 leads across 7 evaluated benchmarks compared to 0 for Claude Haiku 5.5, showing notable advantages in Artificial Analysis Intelligence Index, Artificial Analysis Coding Agent Index v2.0, Terminal-Bench 4.0.
Both models feature equivalent context windows of 1M tokens (1,000,000 tokens).
Claude Haiku 5.5 is more economical. Input tokens cost $0.1/M tokens compared to $4/M tokens for Claude Opus 5.5.
Claude Opus 5.5 is a proprietary cloud API. Claude Haiku 5.5 is a proprietary cloud API.
Analyze 130+ models side-by-side using interactive radar charts, cost-efficiency quadrant plots, and live dataset filters on the full NeuBench index.
Open Interactive Comparator