Codesota · Models · Anthropic Haiku 4.5Anthropic1 results · 1 benchmarks
Model card

Anthropic Haiku 4.5.

AnthropicproprietaryUnknown paramsVision-language model (thinking enabled)

Claude Haiku 4.5 with extended thinking enabled, evaluated as a VLM parser. ParseBench Table 5.

§ 02 · Benchmarks

Every benchmark Anthropic Haiku 4.5 has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01ParseBenchComputer Vision · Document Parsingaccuracy45.2%#13/14—source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 03 · Strengths by area

Where Anthropic Haiku 4.5 actually performs.

Computer Vision
1
benchmark
avg rank #13.0
§ 05 · Related models

Other Anthropic models scored on Codesota.

Claude 3.5 Sonnet
Undisclosed params · 32 results
Claude Opus 4
Undisclosed params · 23 results
Claude Sonnet 4
21 results
Claude Opus 4.6
15 results
Claude 3 Opus
14 results
Claude Sonnet 4.6
13 results
Claude Opus 4.5
Undisclosed params · 12 results
Claude 3.7 Sonnet
10 results
§ 06 · Sources & freshness

Where these numbers come from.

blog-post
1
result
1 of 1 rows marked verified.