Language modelActive

Claude Haiku 4.5

Anthropic

Released: October 2025
Data date: September 8, 2026
Predecessor: Claude Haiku 3.5 (Official documentation)

Speed: 80.6 tokens/s
Output throughput
Price: $1 · $5
Input · Output / 1M tokens
Input: Text, Image
Output: Text

What was new in Claude Haiku 4.5 was Anthropic’s reported coding performance near the Sonnet 4 of that period, at one-third the cost and more than twice the speed. Anthropic released it in October 2025 as the successor to Haiku 3.5.

Anthropic describes a specific division of work with Claude Sonnet 4.5. Sonnet creates the multi-step plan, while several Haiku 4.5 instances complete the subtasks in parallel.

Specifications and access

Specification	Value and source
Weights	ProprietarySource
Access	APISource
Context window	200,000 tokensSource
Knowledge cutoff	February 2025Source
Notes	200K context, 64K output, parameter count not disclosedSource
API model ID	claude-haiku-4-5-20251001Source
Source status	activeSource
Release	2025-10-15Source
Maximum output	64,000 tokensSource
Input	Text, imagesSource
Output	TextSource
Reasoning mode	extended thinking (manual)Source
Knowledge cutoff	2025-02Source
Tool calling	YesSource
Structured output	YesSource

Published benchmarks

CorpFin v2

Overall

1.Claude Opus 573.19%

2.Claude Fable 571.83%

3.Kimi K371.56%

4.Inkling Small69.62%

5.Inkling68.57%

6.Grok 4.368.53%

7.GPT-5.568.42%

8.Kimi K2.568.26%

9.MiniMax M368.1%

10.Grok 4.567.41%

11.Claude Sonnet 566.98%

12.Kimi K2.666.74%

13.Claude Opus 4.866.71%

14.Gemini 3 Flash Preview66.43%

15.Grok 4.666.16%

16.GLM-5.266.12%

17.Claude Opus 4.766.08%

18.Qwen 3.8 Max 090265.85%

19.Nemotron 3 Ultra65.46%

20.Claude Sonnet 4.665.31%

20.GPT-5.6 Terra65.31%

22.GPT-5.465.27%

23.Gemini 3.5 Flash64.69%

24.Gemini 3.1 Pro Preview64.49%

25.GLM-5.164.45%

26.GPT-5.6 Sol64.38%

27.GPT-5.6 Luna64.22%

28.Qwen 3.7 Max63.71%

29.Grok 4.20 Reasoning63.68%

30.Gemini 3.6 Flash63.33%

31.GPT-4.163.05%

32.GLM-562.9%

33.Qwen 3.6 27B62.32%

34.Ling 3.0 Flash61.93%

35.MiMo-V2.5-Pro61.42%

36.DeepSeek-V4-Pro61.38%

37.Claude Opus 4.561.31%

38.GPT-5.4 nano61.19%

38.MiniMax M2.761.19%

40.GPT-561.07%

41.Mistral Large 361.03%

42.GLM-4.560.96%

43.GPT-5.4 mini60.92%

44.Claude Sonnet 4.560.8%

44.Gemini 2.5 Pro60.8%

46.Gemini 3.5 Flash-Lite60.68%

47.Kimi K2 Thinking60.57%

48.Claude Haiku 4.560.3%

49.GPT-5 mini60.18%

50.MiMo-V2.559.91%

51.o359.71%

52.MiniMax M2.559.6%

53.Gemini 3.1 Flash-Lite59.36%

54.o4-mini58.97%

55.MiniMax M2.158.9%

56.Mistral Medium 3.558.78%

57.GPT OSS 120B58.24%

58.GPT-4.1 mini57.93%

59.GLM-4.656.84%

60.DeepSeek-V3 032454.74%

61.Trinity Large Thinking54.66%

62.Nemotron 3.5 Lightning54.24%

63.DeepSeek-R154.12%

64.GPT OSS 20B53.15%

65.DeepSeek-V352.49%

66.DeepSeek-V3.151.48%

67.Kimi K2 071150.39%

68.DeepSeek-V3.247.94%

69.Llama 4 Scout46.78%

70.GLM-4.746.39%

71.Command A45.96%

72.GPT-4o mini45.46%

73.o3-mini45.3%

74.Mistral Small 3.144.17%

75.GPT-4.1 nano42.08%

higher is betterAs of September 29, 2026Scale: 0% - 100%

Source: Vals AI

CC BY 4.0

gradually.ai

Methodology and context

Comparison cohort: vals-corp-fin-v2:2:overall
Metric: accuracy
Scale: 0-100
Participants: 75
Methodology and context: Source model ID: anthropic/claude-haiku-4-5-20251001. Source model name: Claude Haiku 4.5. Model snapshot: claude-haiku-4-5-20251001. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 7.951. Cost per task in USD: 0.075779. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-09-29-83186cfafe52. Source payload hash: 83186cfafe526f0b8fe2ae86af05e8209fe799c8bfa48292ef1fef5db66650ee. Parser version: vals-astro-v2. Imported at: 2026-09-29

Source: Vals AI

Public Benefits Bench v1.1

Overall

1.Claude Opus 576.93%

2.Claude Fable 5.174.9%

3.Claude Opus 5.570.64%

4.Claude Fable 570.43%

5.GLM-5.368.54%

6.Kimi K368.27%

7.Claude Opus 4.868.13%

8.Claude Sonnet 5.567.19%

9.Qwen 3.8 Max 090267.12%

10.Grok 4.666.85%

11.GPT-5.6 Sol66.51%

12.Claude Sonnet 566.04%

13.Grok 4.765.63%

14.Gemini 3.8 Flash65.29%

15.DeepSeek-V4.1-Flash64.28%

16.MiniMax M364.14%

17.DeepSeek-V4-Pro62.92%

18.Claude Sonnet 4.662.45%

19.GPT-5.6 Terra62.38%

20.GLM-5.161.84%

21.GPT-5.6 Luna61.16%

22.GPT-5.560.89%

23.Gemini 3.5 Flash59.47%

24.Inkling Small59.41%

25.Inkling58.46%

26.GPT-6 Luna57.65%

27.GPT-6 Sol56.63%

27.Kimi K2.656.63%

29.Gemini 3.6 Flash56.56%

30.Claude Haiku 4.554.26%

31.Gemini 3.1 Pro Preview53.79%

32.Gemini 3.5 Flash-Lite51.83%

33.Grok 4.351.69%

34.Ling 3.0 Flash50.74%

35.Mercury 2.545.26%

higher is betterAs of September 29, 2026Scale: 0% - 100%

Source: Vals AI

CC BY 4.0

gradually.ai

Methodology and context

Comparison cohort: vals-snap:1.1:overall
Metric: accuracy
Scale: 0-100
Participants: 35
Methodology and context: Source model ID: anthropic/claude-haiku-4-5-20251001. Source model name: Claude Haiku 4.5. Model snapshot: claude-haiku-4-5-20251001. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-09-27. Latency in seconds: 337.201. Cost per task in USD: 0.224512. Source ID: vals-ai. Evidence capture ID: vals-public-benefits-bench-2026-09-29-1769cc3fc7cd. Source payload hash: 1769cc3fc7cd8af239e08dbb9dcf2a08a7bd9d7604016931e6dc7043bde60b1f. Parser version: vals-astro-v2. Imported at: 2026-09-29

Source: Vals AI

Measurements without matching peer values

These measurements have no matching peer values under the same test conditions. Their original values and sources remain available here.

Public Benefits Bench v1: Overall (Show measurement, test conditions, and source)

Source value: 49.526
Score: 49.526
Metric: accuracy
Unit: %
Benchmark version: 1
Category: productivity
Task: overall
Task label: Overall
Direction: Higher is better
Comparison cohort: vals-snap-v1:1:overall
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-09-29
Methodology: Source model ID: anthropic/claude-haiku-4-5-20251001. Source model name: Claude Haiku 4.5. Model snapshot: claude-haiku-4-5-20251001. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-06-09. Latency in seconds: 190.393. Cost per task in USD: 0.194641. Source ID: vals-ai. Evidence capture ID: vals-public-benefits-bench-v1-2026-09-29-659d3d525d58. Source payload hash: 659d3d525d582a0f9b0a59bcfaa2dbbd6b5b20ecb1c6cb4d224f43595a3cc82f. Parser version: vals-astro-v2. Imported at: 2026-09-29
Context: The source dataset marks this benchmark as archived. The result remains visible as historical evidence and is not interchangeable with an active leaderboard.

Source Vals AI

SWE-bench Verified (Show measurement, test conditions, and source)

Source value: 73.3
Score: 73.3
Metric: percent
Unit: %
Category: coding
Direction: Higher is better
Source type: vendor-reported
Evaluator: Anthropic
Status: active
Retrieved at: 2026-09-08
Methodology: Vendor-reported result. 50 trials; no test-time compute; 128K thinking budget; Claude Code harness

Source Anthropic

ScienceAgentBench: Accuracy (Show measurement, test conditions, and source)

Source value: 18.63
Score: 18.63
Metric: accuracy
Unit: %
Benchmark version: verified 102-task set
Category: coding
Task: overall
Task label: Accuracy
Direction: Higher is better
Comparison cohort: hal-scienceagentbench-sab-self-debug-overall
Source type: independent-evaluator
Evaluator: Princeton HAL
Status: active
Retrieved at: 2026-08-06
Methodology: SAB Self-Debug on 102 scientific tasks. Total cost is divided by 102 and excludes caching benefits according to HAL.. Source model ID: curated/claude-haiku-4.5. Source model name: Claude Haiku 4.5. Provider reported by source: Anthropic. Evaluator: Princeton HAL. Dataset type: public. Harness: SAB Self-Debug. Agent: SAB Self-Debug. Cost per task in USD: 0.026078431372549022. Source ID: princeton-hal. Evidence capture ID: curated-princeton-hal-2026-08-06-a3883fc028e8. Source payload hash: a3883fc028e8d720a58b6a0b13e109a42224582734e5e0cc2d89692fdfc3e7e8. Parser version: curated-v1. Imported at: 2026-09-29
Comparison methodology: SAB Self-Debug on 102 scientific tasks. Total cost is divided by 102 and excludes caching benefits according to HAL.

Source Princeton HAL

ScienceAgentBench: Accuracy (Show measurement, test conditions, and source)

Source value: 23.53
Score: 23.53
Metric: accuracy
Unit: %
Benchmark version: verified 102-task set
Category: coding
Task: overall
Task label: Accuracy
Direction: Higher is better
Comparison cohort: hal-scienceagentbench-sab-self-debug-overall
Source type: independent-evaluator
Evaluator: Princeton HAL
Status: active
Retrieved at: 2026-08-06
Methodology: SAB Self-Debug on 102 scientific tasks. Total cost is divided by 102 and excludes caching benefits according to HAL.. Source model ID: curated/claude-haiku-4.5. Source model name: Claude Haiku 4.5. Provider reported by source: Anthropic. Evaluator: Princeton HAL. Dataset type: public. Harness: SAB Self-Debug. Agent: SAB Self-Debug. Reasoning effort: high. Cost per task in USD: 0.03343137254901961. Source ID: princeton-hal. Evidence capture ID: curated-princeton-hal-2026-08-06-b8e1992f26ca. Source payload hash: b8e1992f26ca2a0d85bb2ede2e4f0587814748a4cf78387a29fdc93827babc2e. Parser version: curated-v1. Imported at: 2026-09-29
Comparison methodology: SAB Self-Debug on 102 scientific tasks. Total cost is divided by 102 and excludes caching benefits according to HAL.

Source Princeton HAL

SWE-bench Verified Mini: Accuracy (Show measurement, test conditions, and source)

Source value: 24
Score: 24
Metric: accuracy
Unit: %
Benchmark version: 50-task random verified subset
Category: coding
Task: overall
Task label: Accuracy
Direction: Higher is better
Comparison cohort: hal-swebench-verified-mini-hal-generalist-overall
Source type: independent-evaluator
Evaluator: Princeton HAL
Status: active
Retrieved at: 2026-08-05
Methodology: HAL Generalist on the same random 50-task subset. Total cost is divided by 50.. Source model ID: curated/claude-haiku-4.5. Source model name: Claude Haiku 4.5. Provider reported by source: Anthropic. Evaluator: Princeton HAL. Dataset type: public. Harness: HAL Generalist. Agent: HAL Generalist. Cost per task in USD: 2.9577999999999998. Source ID: princeton-hal. Evidence capture ID: curated-princeton-hal-2026-08-05-f4e1873ff630. Source payload hash: f4e1873ff63084397980aa3b4270ec5733876ce363ba2be214ee709aa6e6604b. Parser version: curated-v1. Imported at: 2026-09-29
Comparison methodology: HAL Generalist on the same random 50-task subset. Total cost is divided by 50.
Context: HAL reports min-max ranges for repeated runs. The central value is the reported mean, not a 95% confidence interval.

Source Princeton HAL

SWE-bench Verified Mini: Accuracy (Show measurement, test conditions, and source)

Source value: 44
Score: 44
Metric: accuracy
Unit: %
Benchmark version: 50-task random verified subset
Category: coding
Task: overall
Task label: Accuracy
Direction: Higher is better
Comparison cohort: hal-swebench-verified-mini-hal-generalist-overall
Source type: independent-evaluator
Evaluator: Princeton HAL
Status: active
Retrieved at: 2026-08-05
Methodology: HAL Generalist on the same random 50-task subset. Total cost is divided by 50.. Source model ID: curated/claude-haiku-4.5. Source model name: Claude Haiku 4.5. Provider reported by source: Anthropic. Evaluator: Princeton HAL. Dataset type: public. Harness: HAL Generalist. Agent: HAL Generalist. Reasoning effort: high. Cost per task in USD: 1.3062. Source ID: princeton-hal. Evidence capture ID: curated-princeton-hal-2026-08-05-92652179b447. Source payload hash: 92652179b4478466a6197e3257bda3003498517dd76e453370cafa7c4a14a68e. Parser version: curated-v1. Imported at: 2026-09-29
Comparison methodology: HAL Generalist on the same random 50-task subset. Total cost is divided by 50.
Context: HAL reports min-max ranges for repeated runs. The central value is the reported mean, not a 95% confidence interval.

Source Princeton HAL

Artificial Analysis Intelligence Index (Show measurement, test conditions, and source)

Source value: 15.410949885381
Score: 15.410949885381
Metric: points
Unit: points
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. Non-reasoning

Source Artificial Analysis

AA-Omniscience Index (Show measurement, test conditions, and source)

Source value: -7.56666666666667
Score: -7.56666666666667
Metric: points
Unit: points
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. Non-reasoning

Source Artificial Analysis

Output speed (Show measurement, test conditions, and source)

Source value: 80.5810327706623
Score: 80.5810327706623
Metric: tokens/s
Unit: tokens/s
Direction: Higher is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. Non-reasoning
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: first-party API.

Source Artificial Analysis

Latency reasoning time (Show measurement, test conditions, and source)

Source value: 0
Score: 0
Metric: seconds
Unit: seconds
Direction: Lower is better
Source type: independent-evaluator
Evaluator: Artificial Analysis
Status: active
Retrieved at: 2026-09-08
Methodology: Independent evaluation. Non-reasoning
Context: Artificial Analysis performance timing; output speed is decoded after the first streaming chunk. Provider basis: first-party API.

Source Artificial Analysis

Recorded measurement details

CorpFin v2: Overall (Show measurement, test conditions, and source)

Source value: 60.295
Score: 60.295
Metric: accuracy
Unit: %
Benchmark version: 2
Category: finance
Task: overall
Task label: Overall
Direction: Higher is better
Comparison cohort: vals-corp-fin-v2:2:overall
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-09-29
Methodology: Source model ID: anthropic/claude-haiku-4-5-20251001. Source model name: Claude Haiku 4.5. Model snapshot: claude-haiku-4-5-20251001. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-08-12. Latency in seconds: 7.951. Cost per task in USD: 0.075779. Source ID: vals-ai. Evidence capture ID: vals-corp_fin_v2-2026-09-29-83186cfafe52. Source payload hash: 83186cfafe526f0b8fe2ae86af05e8209fe799c8bfa48292ef1fef5db66650ee. Parser version: vals-astro-v2. Imported at: 2026-09-29

Source Vals AI

Public Benefits Bench v1.1: Overall (Show measurement, test conditions, and source)

Source value: 54.263
Score: 54.263
Metric: accuracy
Unit: %
Benchmark version: 1.1
Category: productivity
Task: overall
Task label: Overall
Direction: Higher is better
Comparison cohort: vals-snap:1.1:overall
Source type: independent-evaluator
Evaluator: Vals AI
Status: active
Retrieved at: 2026-09-29
Methodology: Source model ID: anthropic/claude-haiku-4-5-20251001. Source model name: Claude Haiku 4.5. Model snapshot: claude-haiku-4-5-20251001. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-09-27. Latency in seconds: 337.201. Cost per task in USD: 0.224512. Source ID: vals-ai. Evidence capture ID: vals-public-benefits-bench-2026-09-29-1769cc3fc7cd. Source payload hash: 1769cc3fc7cd8af239e08dbb9dcf2a08a7bd9d7604016931e6dc7043bde60b1f. Parser version: vals-astro-v2. Imported at: 2026-09-29

Source Vals AI

Page 1 of 2

Predecessor and successor

Sources and data date

Type	Evidence and data date
Speed	Anthropic, Non-reasoning. Output speed after the first streaming chunk, not total response time. Retrieved September 8, 2026. Source: Artificial Analysis · Artificial Analysis (retrieved September 8, 2026)Speed measurement methodology
Price	Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs. Source: Pricing · API pricing
Input	Documented modalities. Retrieved September 8, 2026. Source: Model documentation · Model metadata (retrieved September 8, 2026) · Claude Haiku 4.5 model overview · Editorial description reviewed October 4, 2026
Output	Documented modalities. Retrieved September 8, 2026. Source: Model documentation · Model metadata (retrieved September 8, 2026) · Claude Haiku 4.5 model overview · Editorial description reviewed October 4, 2026
Research status	Research date September 8, 2026. 3 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Not found in the inspected sources	parameters-billions Original wording Anthropic does not publish parameter counts for this model.
Not found in the inspected sources	active-parameters-billions Original wording Anthropic does not publish active parameter counts for this model.
Not found in the inspected sources	profile-benchmarks-additional Original wording The launch page has additional chart material but no other extractable exact rows used here.
Additional source	Model metadata
Additional source	Princeton HAL (retrieved August 6, 2026)
Additional source	Princeton HAL (retrieved August 5, 2026)
Additional source	Vals AI (retrieved September 29, 2026)
Additional source	Vals AI (retrieved September 29, 2026)
Additional source	Vals AI (retrieved September 29, 2026)
Additional source	Anthropic (retrieved September 8, 2026) · Anthropic, Introducing Claude Haiku 4.5 · Editorial description reviewed October 4, 2026
Additional source	Claude Haiku 4.5 system card · Editorial description reviewed October 4, 2026
Additional source	OpenRouter: Popularity index (derived), CC BY 4.0
Additional source	Vercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional source	Creative Commons Attribution 4.0
Additional source	Anthropic Claude API model deprecations page (2026-09-19)

Claude Haiku 4.5

Specifications and access

Published benchmarks

CorpFin v2

Public Benefits Bench v1.1

Measurements without matching peer values

Recorded measurement details

Our tests

Pelican on a bicycle

Popularity

Published prices

Compare this model

Predecessor and successor

Sources and data date