Claude Haiku 5.5 vs Claude Sonnet 5.5
Claude Sonnet 5.5 leads Claude Haiku 5.5 by 11.5 points on the Shogo Agent Index, but their ranges overlap, so treat it as close. Claude Haiku 5.5 wins 0 and Claude Sonnet 5.5 wins 22 of 22 shared benchmarks. Claude Haiku 5.5 is cheaper for agent workloads.
Add more models to this comparison →Pillars
| Pillar | Claude Haiku 5.5 | Claude Sonnet 5.5 |
|---|---|---|
| Agentic coding | 67 | 79 |
| Long-horizon autonomy | 62 | 75 |
| Knowledge-work agents | 46 | 57 |
| Computer use | 47 | 61 |
| Security & context learning | 55 | 61 |
Shared benchmarks
| Benchmark | Claude Haiku 5.5 | Claude Sonnet 5.5 |
|---|---|---|
| APEX-Agentsindex | 56.1% | 75.5% |
| Vending-Bench 2index | $3,602 | $11,165 |
| FrontierSWEindex | 43.8% | 61.9% |
| CursorBenchindex | 48.4% | 55.5% |
| ARC-AGI | 89.0% | 96.5% |
| ARC-AGI-2 | 61.3% | 83.9% |
| Chess Puzzles | 55.0% | 59.0% |
| CritPt | 22.6% | 31.4% |
| DTBench | 95.0% | 97.3% |
| FrontierCode | 46.4% | 52.1% |
| FrontierMath Tier 4 | 46.3% | 80.5% |
| FrontierMath Tiers 1-3 | 75.1% | 88.8% |
| GPQA Diamond | 89.6% | 95.6% |
| Humanity's Last Exam | 30.1% | 46.1% |
| LMCA | 51.1% | 68.3% |
| MirrorCode | 78.9% | 97.8% |
| OTIS Mock AIME | 98.9% | 100.0% |
| ProofBench | 89.0% | 100.0% |
| SciCode | 55.0% | 61.0% |
| SimpleBench | 59.2% | 75.9% |
| SimpleQA Verified | 23.8% | 46.5% |
| WeirdML v3 | 11.7% | 19.9% |
Specs and price
| Claude Haiku 5.5 | Claude Sonnet 5.5 | |
|---|---|---|
| Agent Index | 57.5 (#18) | 69 (#9) |
| Agent $/M | $0.050 | $0.84 |
| Input / output $/M | $0.10 / $0.50 | $2.00 / $10.00 |
| Context | 1M | 1M |
| Released | Oct 7, 2026 | Sep 28, 2026 |
| Weights | Closed | Closed |
| API hosts | 5 | 5 |
Don’t pick one. Route.
Shogo can run Claude Haiku 5.5 and Claude Sonnet 5.5 side by side and send each step of an agent to the one that’s cheaper for it.
Try both in Shogo