Claude Haiku 4.5 (Thinking 32K): benchmark scores, price and ranking

Claude Haiku 4.5 (Thinking 32K) from Anthropic ranks #412 of 743 LLMs on the LLMs Tiger aggregate benchmark score (47.7), based on 1 benchmark results captured up to 2026-10-13.

Organisation: Anthropic
Type: Fast / low-cost (flash) model
Open weights: No
Released: 2025-10-15 (source: curated)
Cheapest API price: $1.00 in / $5.00 out per million tokens (via anthropic, observed 2026-10-13T20:47Z)
Aggregate rank: #412 of 743 (score 47.7)

Across the 1 benchmark where at least ten models are scored, Claude Haiku 4.5 (Thinking 32K) is in the top 10% on 0 and in the top half on 0.

Its cheapest listed input price, $1.00 per million tokens, is lower than that of 35% of the 427 models with a listed price.

Claude Haiku 4.5 (Thinking 32K) benchmark results

Claude Haiku 4.5 (Thinking 32K) benchmark scores, ranked against all models measured on each benchmark
Benchmark	Score	Rank	Source	Captured
ARC-AGI-2	4.0%	#62 of 87	ARC Prize official leaderboard	2026-10-13

About Claude Haiku 4.5 (Thinking 32K)

Claude Haiku 4.5 (Thinking 32K) is a smaller, lower-cost model from Anthropic. It was released on 2025-10-15, according to the developer's announcement. The cheapest API listing we track charges $1.00 per million input tokens and $5.00 per million output tokens, through anthropic. We hold 1 benchmark result for it across 1 area, and none is in the top tenth of its board.

Where Claude Haiku 4.5 (Thinking 32K) stands

On our aggregate score, Claude Haiku 4.5 (Thinking 32K) is #412 of 743 ranked models, around the middle of the table. Its score is 47.7; the scale is centred on 50, so it sits 2.3 points below the typical model. That beats 45% of the ranked field. Among the 70 ranked Anthropic models it is #53. The next one up is claude-3-7-sonnet-20250219 at 48.2. The next one down is Claude 3.7 at 47.4.

Claude Haiku 4.5 (Thinking 32K), benchmark by benchmark

Claude Haiku 4.5 (Thinking 32K) lands below the middle on ARC-AGI-2: 4.0%, #62 of 87, with 70% of the field scoring higher. GPT-6 Astra (max) leads at 95.0%; the closest result above is GPT-5 mini (high) at 4.4%. It covers abstract reasoning, ARC Prize public eval set. The board median is 22.8%, so Claude Haiku 4.5 (Thinking 32K) is 18.8 points below it. Source: ARC Prize official leaderboard, captured 2026-10-13.

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so Claude Haiku 4.5 (Thinking 32K) costs more than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4.1 Flash (high) ($0.010, 48.9), Qwen2.5-Coder-3B-Instruct ($0.010, 46.5). Similar scores cost more on o1-mini (high) ($1.10) and mimo-v2-pro ($1.10).

Release history

The Anthropic model released before it was claude-haiku-4-5-20251001_32K on 2025-10-15, which scores 47.3. The next one was Claude 4.5 Opus on 2025-11-24, at 54.5.

Reading these numbers

The 1 score on this page comes from 1 source: ARC Prize official leaderboard (1). 1 of them come from an official leaderboard and 0 from datasets or third-party runs. The leaderboard also lists 6 other reasoning-effort settings of this model, which redirect here. This page shows the highest. With 1 results the aggregate rank can move a lot when one new score arrives.

Claude Haiku 4.5 (Thinking 32K) is also listed at 6 other reasoning-effort settings; this page shows the highest-effort row, which is the one the leaderboard keeps by default.

Models ranked nearby

#409 Gemini 2.5 Pro (Thinking 32K) (47.7)
#410 GLM-4.7 (47.7)
#411 Mistral Large 4 (high) (47.7)
#413 gemini-3.1-flash-lite (high) (47.7)
#414 ollama/codestral (47.6)
#415 Gemini 2.5 Flash (Preview) (Thinking 24K) (47.6)