Rankings / Anthropic
Claude 3 Haiku
released 2023-04-30
BenchAtlas Index
as of 2026-07-14
base configuration
rank #425
10 families · 5 categories · high
Benchmark evidence
12 results
agentic coding
| τ²-Bench | 21.1 % | independent | Artificial Analysis |
coding
| LiveCodeBench v6 problems=454 · window_start=2024-08-01 | 20.2 % | independent | LiveCodeBench Leaderboard | |
| LiveCodeBench v5older version problems=279 · window_start=2024-08-01 | 17.7 % | independent | LiveCodeBench Leaderboard | |
| LiveCodeBench v6 implementation=artificial-analysis | 15.4 % | independent | Artificial Analysis | |
| SciCode | 18.6 % | independent | Artificial Analysis |
external indices
| Intelligence Index v4.1 | 3.9 points | independent | Artificial Analysis |
knowledge science
| GPQA Diamond implementation=artificial-analysis | 37.4 % | independent | Artificial Analysis | |
| Humanity's Last Exam implementation=artificial-analysis | 3.9 % | independent | Artificial Analysis |
long context instruction
| AA-LCR | 21.0 % | independent | Artificial Analysis | |
| IFBench | 36.1 % | independent | Artificial Analysis |
reasoning math
| AIME implementation=artificial-analysis | 1.0 % | independent | Artificial Analysis | |
| MATH-500 implementation=artificial-analysis | 39.4 % | independent | Artificial Analysis |
Agent + model results
systems, not bare-model scores
| agent + model Artificial Analysis harness + Claude 3 Haiku | Terminal-Bench Hard | 0.8 % | independent | Artificial Analysis |
These scores measure the whole agent system (scaffold, tools, budgets) — they are never merged into the bare model’s numbers.