Skip to main content

AnthropicLanguage modelActive

Claude Haiku 5.5

Claude Haiku 5.5 is among the strongest models: Rank 20 of 91 in the overall index. Through the API it is cheap.

IntelligenceRank 20 of 91

103.4Overall indexhigher is better

SpeedRank 5 of 100

236Tokens per secondhigher is better

PriceRank 8 of 94

Input $0.10 · Output $0.50

$0.20per 1M tokenslower is better

ContextRank 31 of 101

1Mtokens at once≈ 1,500 A4 pages

PopularityRank 3 of 875

77out of 100same as last month

Summary

It is cheap: $0.20 per 1M tokens (input and output blended); a typical model costs $1.56. At 236 tokens per second it writes faster than a typical model (87).

What is new in Claude Haiku 5.5 is that Anthropic offers adjustable effort on a Haiku model for the first time. It follows Claude Haiku 4.5 and grows the context window from 200,000 to 1 million tokens. Anthropic positions it for classification, routing, extraction, and subagents, wherever many requests need fast answers.

Its pricing by prompt length is unusual. Up to 100,000 tokens, Anthropic puts the rate about 90% below Haiku 4.5; above that, the whole request pays five times as much. Because the new tokenizer splits the same text into roughly 30% more tokens, real savings are smaller. Anthropic reports about 75% lower cost on average.

Claude Haiku 5.5 benchmarks

Overall index (Gradually Meta-Score v4)

20 of 91 models

4 models in this view are measured at a lower reasoning effort than max in the overall index. Dashed column: lower reasoning effort than the top.

Claude Opus 5.5max126.1
Gemini 4 Carbonhigh125.7
Claude Fable 5.1max123.8
Claude Sonnet 5.5max122.8
Gemini 4 Argonhigh122.3
GPT-6 Astramax121.3
GPT-6.1 Solmax119.2
Muse Wonder Fullmax118.3
Claude Opus 5max117.2
Muse Wondermax113.4
Kimi K3.1 Promax112.9
Claude Fable 5max112.5
GPT-6 Solmax110.6
GPT-5.6 Solmax108.7
Muse Spark 1.3max107.4
Gemini 3.7 Flashhigh107.1
DeepSeek V4.1 Flash 1017max106.6
Gemini 3.8 Flashhigh106.5
GPT-6.1 Terramax105.2
Claude Haiku 5.5max103.4

Claude Haiku 5.5 by reasoning effort

Points on the overall index scale. The value at max is Claude Haiku 5.5's overall index; the other efforts derive from the same Artificial Analysis test series. Missing efforts are never estimated.

7090110130nonelowmediumhighxhighmaxGrok 4.7 · low: 98.7Grok 4.7 · high: 103.6Grok 4.7 · xhigh: 103.3GPT-6.1 Terra · none: 86.1GPT-6.1 Terra · low: 89.4GPT-6.1 Terra · medium: 97.1GPT-6.1 Terra · high: 99.5GPT-6.1 Terra · xhigh: 101.9GPT-6.1 Terra · max: 105.2Claude Haiku 5.5 · low: 80.8Claude Haiku 5.5 · medium: 90.4Claude Haiku 5.5 · high: 96.6Claude Haiku 5.5 · xhigh: 101.3Claude Haiku 5.5 · max: 103.4
Claude Haiku 5.5low 80.8 · medium 90.4 · high 96.6 · xhigh 101.3 · max 103.4Grok 4.7low 98.7 · high 103.6 · xhigh 103.3GPT-6.1 Terranone 86.1 · low 89.4 · medium 97.1 · high 99.5 · xhigh 101.9 · max 105.2Filled dot: effort of the overall index. Shading: 95% range.

Source: Meta-Score v4; tiers: Artificial Analysis

Speed

20 of 100 models

Mercury 2.5576tokens/s · Configuration Reasoning enabled; effort unspecified · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Gemini 3.1 Flash-Lite336tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Gemini 3.5 Flash-Lite299tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Nemotron 3.5 Lightning245tokens/s · Configuration Reasoning enabled; effort unspecified · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Claude Haiku 5.5236tokens/s · Configuration Max effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
o3-mini230tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Ling 3.0 Flash199tokens/s · Configuration Reasoning enabled; effort unspecified · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
DeepSeek-V3 0324195tokens/s · Configuration Non-reasoning · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
GPT-5.4 mini191tokens/s · Configuration Xhigh effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Gemini 3 Flash Preview191tokens/s · Configuration Non-reasoning · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Gemini 3.5 Flash184tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Gemini 3.6 Flash182tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Command A Plus177tokens/s · Configuration Reasoning enabled; effort unspecified · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
DeepSeek-V4.1-Flash176tokens/s · Configuration Max effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
DeepSeek V4.1 Flash 1017172tokens/s · Configuration Max effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Qwen 3.7 Max165tokens/s · Configuration Reasoning enabled; effort unspecified · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
GPT-5.4 nano153tokens/s · Configuration Xhigh effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
GPT OSS 20B152tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
GPT OSS 120B151tokens/s · Configuration High effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06
Claude Sonnet 5.5151tokens/s · Configuration Max effort · AA 3.0.0, including time to first token, averaged across available workloads and five days · Retrieved 2026-11-06

Output tokens per second, measured by Artificial Analysis

Intelligence vs. price

most attractive area: stronger and cheaper than half of the models shownbest intelligence for the price

80100120$0.10$0.50$2$10$20Gemini 3.6 Flash: 96.7, $1.50Step 5 Preview: 97.3, $1.43MiMo V2.6 Flash: 98.3, $0.18Claude Opus 4.8: 99.0, $10DeepSeek-V4.1-Flash: 99.7, $0.26Qwen 3.8 Max 0902: 100.3, $2.48GLM-5.3: 100.7, $2.15GPT-5.6 Terra: 100.7, $4.50MiMo V2.6 Pro: 100.9, $0.54GPT-5.5: 100.9, $11.25Kimi K3: 101.2, $6Grok 4.6: 102.4, $3Grok 4.7: 103.3, $3GPT-6.1 Terra: 105.2, $2Gemini 3.8 Flash: 106.5, $1.50DeepSeek V4.1 Flash 1017: 106.6, $0.26Muse Spark 1.3: 107.4, $2GPT-5.6 Sol: 108.7, $8GPT-6 Sol: 110.6, $4Claude Fable 5: 112.5, $20Kimi K3.1 Pro: 112.9, $4Muse Wonder: 113.4, $8Claude Opus 5: 117.2, $10GPT-6.1 Sol: 119.2, $4GPT-6 Astra: 121.3, $20Gemini 4 Argon: 122.3, $4Claude Sonnet 5.5: 122.8, $4Claude Fable 5.1: 123.8, $20Claude Opus 5.5: 126.1, $8Claude Haiku 5.5: 103.4, $0.20Claude Haiku 5.5Claude Opus 5.5Claude Fable 5.1Claude Sonnet 5.5Gemini 4 ArgonGPT-6 AstraGPT-6.1 SolClaude Opus 5Muse WonderKimi K3.1 ProClaude Fable 5GPT-6 SolGPT-5.6 SolMuse Spark 1.3DeepSeek V4.1 Flash 1017Gemini 3.8 FlashGPT-6.1 TerraGrok 4.7Grok 4.6Kimi K3GPT-5.5MiMo V2.6 ProGPT-5.6 TerraGLM-5.3Qwen 3.8 Max 0902DeepSeek-V4.1-FlashClaude Opus 4.8MiMo V2.6 FlashStep 5 PreviewGemini 3.6 FlashPrice per 1M tokens (blended)Overall index

Claude Haiku 5.5 by task area

Points, 100 = reference group average · higher is better

HealthRank 10 of 68108.9
MathematicsRank 8 of 44107.1
GeneralRank 13 of 60106.4
CodingRank 18 of 90106.2
AgentsRank 19 of 83106.1
ReasoningRank 18 of 58105.2
FinanceRank 22 of 96104.7
LegalRank 38 of 6694.4
KnowledgeRank 23 of 3185.5

No tests yet: Multimodal.

Text Arena

How people rate the answers in blind comparisons, Elo points with style control · higher is better

20 of 40 models

Gemini 4 Argon1,520
Claude Opus 5.51,509
Muse Wonder1,507
Claude Fable 5.11,504
Claude Fable 51,504
Gemini 3.8 Flash1,496
Muse Spark 1.31,495
Claude Opus 51,491
Gemini 3.7 Flash1,490
Kimi K31,488
Gemini 3.1 Pro Preview1,487
GPT-5.6 Sol1,485
Gemini 3.6 Flash1,483
GPT-6.1 Sol1,481
GLM-5.31,481
Gemini 3.5 Flash1,478
Claude Sonnet 5.51,477
GLM-5.3-Flash1,477
DeepSeek-V4.1-Flash1,476
Claude Haiku 5.51,425

as of October 30, 2026 · Source: LMArena

WebDev Arena

How people rate generated websites and web apps in blind comparisons, Elo points · higher is better

20 of 43 models

Claude Opus 5.51,810
GPT-6 Astra1,773
Claude Sonnet 5.51,770
GPT-6.1 Sol1,751
Claude Fable 5.11,746
Muse Wonder Full1,725
Claude Opus 51,691
Gemini 4 Argon1,679
GPT-6 Sol1,678
Qwen 3.8 Max 09021,676
DeepSeek V4.1 Flash 10171,665
Kimi K31,656
Muse Spark 1.31,656
Muse Wonder1,652
Grok 4.71,637
Claude Fable 51,624
GLM-5.31,623
DeepSeek-V4.1-Flash1,619
Grok 4.61,616
Claude Haiku 5.51,589

as of October 30, 2026 · Source: LMArena

Terminal-Bench 4.0

Share of tasks the model solves as an agent in the terminal, in percent · higher is better

20 of 38 models

Claude Opus 5.565.2 %
Claude Sonnet 5.564.1 %
Kimi K3.1 Pro63.6 %
Gemini 4 Carbon63.1 %
GPT-6 Astra59.6 %
Claude Fable 5.158.1 %
Muse Wonder Full58.1 %
Gemini 4 Argon57.6 %
GPT-6.1 Sol55.1 %
Claude Opus 553.5 %
GPT-6.1 Terra49.5 %
Muse Wonder49.5 %
GPT-6 Sol44.4 %
Claude Fable 541.4 %
GLM-5.338.9 %
GPT-5.6 Sol37.9 %
GLM-5.3-Flash37.4 %
Claude Haiku 5.536.4 %
Qwen 3.8 Max 090234.3 %
DeepSeek V4.1 Flash 101731.8 %

as of November 6, 2026 · Source: Vals AI

Vibe Code Bench

Share of apps the model builds correctly from a description, in percent · higher is better

20 of 51 models

Claude Sonnet 5.592.4 %
Gemini 4 Argon91.9 %
Claude Haiku 5.590.4 %
Claude Fable 590.4 %
Claude Opus 5.590.3 %
Claude Fable 5.190.3 %
GPT-6 Astra89.6 %
Gemini 4 Carbon89.1 %
GPT-6.1 Sol88.9 %
Claude Opus 588.4 %
GPT-6 Sol87.8 %
DeepSeek V4.1 Flash 101787.7 %
Grok 4.786.2 %
Muse Wonder Full86.0 %
Muse Spark 1.385.9 %
Kimi K385.0 %
Kimi K3.1 Pro84.9 %
DeepSeek-V4.1-Flash84.7 %
GPT-6.1 Terra84.0 %
Muse Wonder83.1 %

as of November 6, 2026 · Source: Vals AI

Show all 984 published benchmark results

Page 1 of 82

Published benchmarks

Other tasks

Vals Index

Code Migration v1.1

Claude Sonnet 5.587.37%
Claude Opus 5.580.49%
Claude Haiku 5.550.2%

higher is betterAs of November 6, 2026Scale: 0% - 100%

Source: Vals AI
gradually.ai
Methodology and context
Comparison cohort
vals-vals-index:2.2:code_migration
Metric
benchmark score
Scale
0-100
Participants
3
Methodology and context
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Provider reported by source: Anthropic. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-05. Compute: max. Maximum output tokens: 128000. Latency in seconds: 5148.212. Standard error: 4.208. Cost per task in USD: 6.130933. Source ID: vals-ai. Evidence capture ID: vals-vals_index-2026-11-06-9440acd92c48. Source payload hash: 9440acd92c489f5f417f5b98dccda8dfcb98774e52cd6641e3eb61acec0917d9. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source: Vals AI

Vals Router Index

Sub-taskClaude Haiku 5.5 (This model)Claude Sonnet 5.5GPT-6 AstraGPT-6.1 SolGemini 4 ArgonMuse WonderDeepSeek V4.1 Flash 1017Muse Wonder FullGLM-5.3Grok 4.7GLM-5.3-FlashClaude Opus 5.5GPT-6 LunaGemini 3.8 FlashKimi K3
Code Migration50.2%87.37% (best value in this row)78.03%75.14%74.43%69.78%68.5%61.65%54.08%51.82%50.63%47.49%45.45%38.17%22.67%
Coding59%80.8% (best value in this row)75.74%73.04%74.64%67.46%62.68%68.58%57.03%55.59%39.59%62.97%46.91%45.51%41.6%
EMB65.48%75.9% (best value in this row)72.68%71.09%74.83%72.5%69.17%75%69.11%71.5%66.25%75.84%69.95%72.87%71.26%
Finance Agent65.37%61.7%60.43%61.82%67.66% (best value in this row)65.55%62.93%67.58%65%60.59%60.59%66.76%56.84%65.76%60.01%
HLAB4.17%10.42%7.08%6.25%27.08% (best value in this row)23.75%14.17%23.33%14.58%18.33%10.42%9.17%4.17%12.08%12.92%
Legal Research43.27%44.23%39.42%38.46%54.81% (best value in this row)50.48%45.19%53.37%49.04%47.12%45.19%50.48%30.29%38.94%46.15%
Full Index58.71%69.03%65.02%63.94%69.24% (best value in this row)64.88%61.26%66.79%59.73%58.56%50.86%64.16%52.32%55.95%52.96%
Tax Agent Bench31.92%40.92%20.72%21.75%44.56%36.27%44.35%43%39.7%34.2%31.55%45.57% (best value in this row)19.95%32.12%33.37%
Terminal-Bench 4.036.36%62.63% (best value in this row)59.6%55.05%57.58%49.5%31.82%58.08%38.89%28.79%37.37%58.08%13.64%19.7%17.17%
Vibe Code Bench90.44%92.39% (best value in this row)89.59%88.93%91.91%83.09%87.73%86%78.13%86.18%30.76%83.34%81.65%78.65%84.97%

higher is betterAs of November 6, 2026Source: Vals AIbars scaled within each row, not across rows

Methodology and context
Metric
accuracy

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.

AA-AnalystAgent: Overall (Show measurement, test conditions, and source)
Source value
53.75
Score
53.75
Metric
pass^5 reliability
Unit
%
Benchmark version
AA cohort: 80 questions
Category
agentic
Task
overall
Task label
Overall
Direction
Higher is better
Comparison cohort
artificial-analysis-aa-analyst-agent:AA cohort: 80 questions:standalone:reasoning:tokens=0:suffix=Max
Source type
independent-evaluator
Evaluator
Artificial Analysis
Status
superseded
Retrieved at
2026-10-19
Methodology
Exact AA configuration: Claude Haiku 5.5 (Max) (release claude-haiku-5-5, row claude-haiku-5-5, UUID 7d064a11-6e78-4030-b535-cdf940615041, reasoning=true, source reasoning-token configuration budget=0, suffix=Max). This configured value is not measured token consumption. 80 quantitative questions across 14 domains are run five independent times. pass^5 counts only questions answered correctly in all five attempts.. Source model ID: artificial-analysis/7d064a11-6e78-4030-b535-cdf940615041/claude-haiku-5-5/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Provider reported by source: Anthropic. Model snapshot: Claude Haiku 5.5 (Max). Evaluator: Artificial Analysis. Dataset type: artificial-analysis-evaluation. Harness: Artificial Analysis Stirrup. Agent: Stirrup analyst agent. Tool policy: 100-turn cap; Python sandbox with source files, URL fetch, image viewing, and answer submission. Attempts: pass^5, 5 independent attempts per question. Reasoning: reasoning enabled. Reasoning effort: Max. Score display: 53.8%. Source ID: artificial-analysis. Evidence capture ID: aa-model-page-2026-10-19-8a022369738c. Source payload hash: 8a022369738cbe2a0e2f9bf486a2882865dbb0c6f1396dfec84059a802560e3c. Parser version: artificial-analysis-rsc-models-v2. Imported at: 2026-10-19
Comparison methodology
Exact AA configuration: Claude Haiku 5.5 (Max) (release claude-haiku-5-5, row claude-haiku-5-5, UUID 7d064a11-6e78-4030-b535-cdf940615041, reasoning=true, source reasoning-token configuration budget=0, suffix=Max). This configured value is not measured token consumption. 80 quantitative questions across 14 domains are run five independent times. pass^5 counts only questions answered correctly in all five attempts.. Harness: Artificial Analysis Stirrup. Agent: Stirrup analyst agent. Tool policy: 100-turn cap; Python sandbox with source files, URL fetch, image viewing, and answer submission. Attempts: pass^5, 5 independent attempts per question. Reasoning: reasoning enabled. Reasoning effort: Max
Context
Questions, source files, and reference answers are private. pass^5 measures reliability across all five attempts, not pass@1.. This result is retained as historical evidence and is not interchangeable with current results.
Source Artificial Analysis

Recorded measurement details

Vals Index: Code Migration v1.1 (Show measurement, test conditions, and source)
Source value
50.199
Score
50.199
Metric
benchmark score
Unit
%
Benchmark version
2.2
Category
aggregate
Task
code_migration
Task label
Code Migration v1.1
Direction
Higher is better
Comparison cohort
vals-vals-index:2.2:code_migration
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Provider reported by source: Anthropic. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-05. Compute: max. Maximum output tokens: 128000. Latency in seconds: 5148.212. Standard error: 4.208. Cost per task in USD: 6.130933. Source ID: vals-ai. Evidence capture ID: vals-vals_index-2026-11-06-9440acd92c48. Source payload hash: 9440acd92c489f5f417f5b98dccda8dfcb98774e52cd6641e3eb61acec0917d9. Parser version: vals-astro-v4. Imported at: 2026-11-06
Comparison methodology
Compute: max. Maximum output tokens: 128000
Source Vals AI
Vals Router Index: Code Migration (Show measurement, test conditions, and source)
Source value
50.199
Score
50.199
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
code_migration
Task label
Code Migration
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:code_migration
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 5148.212. Standard error: 4.208. Cost per task in USD: 6.130933. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Coding (Show measurement, test conditions, and source)
Source value
59.00233333333333
Score
59.00233333333333
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
coding
Task label
Coding
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:coding
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 3446.773333333333. Standard error: 1.7008979654549794. Cost per task in USD: 5.106751. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: EMB (Show measurement, test conditions, and source)
Source value
65.477
Score
65.477
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
emb
Task label
EMB
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:emb
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 1195.527. Standard error: 3.691. Cost per task in USD: 1.364559. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Finance Agent (Show measurement, test conditions, and source)
Source value
65.373
Score
65.373
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
finance_agent
Task label
Finance Agent
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:finance_agent
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 1864.634. Standard error: 2.063. Cost per task in USD: 1.637243. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: HLAB (Show measurement, test conditions, and source)
Source value
4.167
Score
4.167
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
legal_agent_benchmark
Task label
HLAB
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:legal_agent_benchmark
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 2620.795. Standard error: 1.518. Cost per task in USD: 4.264376. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Legal Research (Show measurement, test conditions, and source)
Source value
43.269
Score
43.269
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
legal_research
Task label
Legal Research
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:legal_research
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 4911.734. Standard error: 3.444. Cost per task in USD: 1.633459. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Full Index (Show measurement, test conditions, and source)
Source value
58.708
Score
58.708
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
overall
Task label
Full Index
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:overall
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 2988.332. Standard error: 1.282. Cost per task in USD: 3.173012. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Tax Agent Bench (Show measurement, test conditions, and source)
Source value
31.922
Score
31.922
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
tax_agent_bench
Task label
Tax Agent Bench
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:tax_agent_bench
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 2973.647. Standard error: 3.352. Cost per task in USD: 1.164207. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Terminal-Bench 4.0 (Show measurement, test conditions, and source)
Source value
36.364
Score
36.364
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
terminal_bench_4
Task label
Terminal-Bench 4.0
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:terminal_bench_4
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 3116.385. Standard error: 2.314. Cost per task in USD: 3.117464. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI
Vals Router Index: Vibe Code Bench (Show measurement, test conditions, and source)
Source value
90.444
Score
90.444
Metric
accuracy
Unit
%
Benchmark version
1.0
Category
aggregate
Task
vibe_code_bench
Task label
Vibe Code Bench
Direction
Higher is better
Comparison cohort
vals-vals-router-index:1.0:vibe_code_bench
Source type
independent-evaluator
Evaluator
Vals AI
Status
active
Retrieved at
2026-11-06
Methodology
Source model ID: anthropic/claude-haiku-5-5. Source model name: Claude Haiku 5.5. Evaluator: Vals AI. Dataset type: private. Leaderboard date: 2026-11-03. Latency in seconds: 2075.723. Standard error: 1.725. Cost per task in USD: 6.071856. Source ID: vals-ai. Evidence capture ID: vals-vals_router_index-2026-11-06-7259f1a74128. Source payload hash: 7259f1a741284313ec1d3cb9b7764e95c6e80b7cbaa3a10f99483a6c4a404336. Parser version: vals-astro-v4. Imported at: 2026-11-06
Source Vals AI

Page 1 of 82

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

The 100 most popular models of this category, 0 to 100, higher is more popular

20 of 100 models

Gemini 4 Argon92
GPT-6 Astra78
Claude Haiku 5.577
Qwen3.5 4B76
GPT-573
GPT-6 Sol72
Qwen3.5 0.8B72
Qwen3.5 2B71
Qwen 3.8 27B70
Claude Opus 5.570
DeepSeek V4.1 Flash 101768
Qwen 3.8 Max 090268
Kimi K368
DeepSeek-V4-Flash65
GPT-5.6 Terra64
Claude Sonnet 563
DiffusionGemma 26B A4B IT63
Claude Opus 562
Qwen 3.8 Flash62
MiniCPM-V 4.7 35B-A3B62

As of November 2026 · Source: gradually.ai popularity index

Over time

Weekly values from October 18, 2026 to November 5, 2026

Claude Haiku 5.5GPT-6 AstraGPT-6 SolQwen 3.8 27BClaude Opus 5.5

Claude Haiku 5.5 compared

Compare with another model

The most popular comparisons

  1. 1Claude Haiku 5.5 vs. GPT-6 Luna
  2. 2Claude Sonnet 5.5 vs. Claude Haiku 5.5
  3. 3Claude Haiku 5.5 vs. DeepSeek V4.1 Flash 1017
  4. 4Claude Haiku 5.5 vs. DeepSeek-V4-Flash
  5. 5Claude Haiku 5.5 vs. DeepSeek-V4.1-Flash
  6. 6Claude Opus 5.5 vs. Claude Haiku 5.5
  7. 7Claude Haiku 5.5 vs. GLM-5.3
  8. 8Claude Haiku 5.5 vs. Gemini 3.5 Flash-Lite
  9. 9Claude Haiku 5.5 vs. Gemini 3.8 Flash
Show all 56 comparisons
  1. 10Claude Haiku 5.5 vs. Claude Sonnet 4.5
  2. 11Claude Haiku 5.5 vs. Claude Sonnet 5
  3. 12Claude Haiku 5.5 vs. Claude Opus 5
  4. 13Claude Haiku 5.5 vs. GPT-6 Astra
  5. 14Claude Haiku 5.5 vs. Gemini 3.5 Flash
  6. 15Claude Haiku 5.5 vs. Claude Opus 4.8
  7. 16Claude Haiku 5.5 vs. Gemini 4 Argon
  8. 17Claude Haiku 5.5 vs. Grok 4.7
  9. 18Claude Haiku 5.5 vs. GPT-5.6 Luna
  10. 19Claude Haiku 5.5 vs. Qwen 3.8 Max 0902
  11. 20Claude Haiku 5.5 vs. Muse Spark 1.3
  12. 21Claude Haiku 5.5 vs. Claude Sonnet 4.6
  13. 22Claude Haiku 5.5 vs. Gemini 3.7 Flash
  14. 23Claude Haiku 5.5 vs. Mistral Large 4
  15. 24Claude Haiku 5.5 vs. GPT-6.1 Sol
  16. 25Claude Haiku 5.5 vs. Gemini 3.1 Pro Preview
  17. 26Claude Haiku 5.5 vs. GLM-5.3-Flash
  18. 27Claude Haiku 5.5 vs. DeepSeek-V4-Pro
  19. 28Claude Haiku 5.5 vs. GPT-5.6 Sol
  20. 29Claude Haiku 5.5 vs. Gemini 3.6 Flash
  21. 30Claude Haiku 5.5 vs. Claude Opus 4.6
  22. 31Claude Haiku 5.5 vs. Kimi K2.6
  23. 32Claude Haiku 5.5 vs. Claude Opus 4.7
  24. 33Claude Haiku 5.5 vs. Mistral Large 3
  25. 34Claude Haiku 5.5 vs. GPT-5.5
  26. 35Claude Haiku 5.5 vs. GPT-6.1 Terra
  27. 36Claude Haiku 5.5 vs. Claude Fable 5.1
  28. 37Claude Haiku 5.5 vs. Claude Fable 5
  29. 38Claude Haiku 5.5 vs. Claude Opus 4.5
  30. 39Claude Haiku 5.5 vs. GPT-6 Sol
  31. 40Claude Haiku 5.5 vs. GPT-5.6 Terra
  32. 41Claude Haiku 5.5 vs. GPT-5.4
  33. 42Claude Haiku 5.5 vs. GPT-5.4 mini
  34. 43Claude Haiku 5.5 vs. Gemini 4 Carbon
  35. 44Claude Haiku 5.5 vs. Muse Wonder
  36. 45Claude Haiku 5.5 vs. Muse Wonder Full
  37. 46Claude Haiku 5.5 vs. Grok 4.6
  38. 47Claude Haiku 5.5 vs. Grok 4.5
  39. 48Claude Haiku 5.5 vs. Grok 4.3
  40. 49Claude Haiku 5.5 vs. Kimi K3.1 Pro
  41. 50Claude Haiku 5.5 vs. Kimi K3
  42. 51Claude Haiku 5.5 vs. GLM-5.2
  43. 52Claude Haiku 5.5 vs. GLM-5.1
  44. 53Claude Haiku 5.5 vs. Qwen 3.8 27B
  45. 54Claude Haiku 5.5 vs. Mistral Medium 3.5
  46. 55Claude Haiku 5.5 vs. MiMo-V2.5
  47. 56Claude Haiku 5.5 vs. MiMo-V2.5-Pro

All specifications

Architecture

Knowledge cutoff
June 2026
Source: Knowledge cutoff

Input and output

Context window
1,000,000 tokens
Source: Context window
Maximum output
128,000 tokens
Source: Maximum output
Input
Text, images
Source: Input
Output
Text
Source: Output
Reasoning mode
adaptive (on by default); default medium
Source: Reasoning mode

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking
Yes. Anthropic confirms text watermarking for this model across its own surfaces and the listed cloud partners worldwide.
Source 1: AI watermarkingSource 2: AI watermarking
Watermarking technique
Variant of SynthID Text
Source 1: Watermarking techniqueSource 2: Watermarking technique
Provenance metadata
C2PA Content Credentials for supported generated files, where the platform offers Claude file generation.
Source: Provenance metadata
Labelling reviewed
Nov 5, 2026
Source 1: Labelling reviewedSource 2: Labelling reviewed

Licence and access

Weights
Proprietary
Source: Weights
Access
API
Source: Access
Source status
active
Source: Source status
Release
Oct 7, 2026
Source: Release

API

API model ID
claude-haiku-5-5
Source: API model ID

Further details

Notes
Released October 7, 2026. Anthropic positions Haiku 5.5 for high-volume, latency-sensitive work such as classification, routing, extraction, and subagent tasks. Pricing depends on prompt length: $0.10 / $0.50 per million input and output tokens for prompts up to 100,000 tokens, $0.50 / $2.50 above that. 1M context, 128K output, a June 2026 knowledge cutoff, adaptive thinking on by default, and medium effort by default. Official API ID: claude-haiku-5-5.
Source: Notes

Prices

Choose a price view

11 offers, prices in US dollars per 1M tokens, blended from 3 parts input and 1 part output.

Filter offers
ProviderInputOutputCache readCache writeBlended
Anthropicdirect$0.10$0.50$0.01$0.125$0.20
Anthropicabove 100,000 context tokensdirect$0.50$2.50$0.05$0.625$1
Amazon Bedrock$0.10$0.50n/an/a$0.20
Anthropic$0.10$0.50n/an/a$0.20
AzureGlobal$0.10$0.50n/an/a$0.20
Claude Platform on AWS$0.10$0.50n/an/a$0.20
GoogleGlobal$0.10$0.50n/an/a$0.20
Amazon BedrockEu-west-1$0.11$0.55n/an/a$0.22
Amazon BedrockUs-east-1$0.11$0.55n/an/a$0.22
AzureUS$0.11$0.55n/an/a$0.22
GoogleEurope$0.11$0.55n/an/a$0.22
GoogleUS$0.11$0.55n/an/a$0.22

Cache prices come from the direct tariff; OpenRouter does not report them for the other providers. Direct price according to the developer, checked Oct 13, 2026. Provider prices via OpenRouter, as of Nov 6, 2026. Open the LLM price index

More price tiers and rules
Cache write, 1 hour
$0.2 per 1M tokensSource
Cache write, 1 hour
$1 per 1M tokens (above 100,000 context tokens)Source
Source pricing conditions
Prompt-length pricing: prompts over 100,000 tokens (including cache reads and writes) pay $0.50 input / $2.50 output instead of $0.10 / $0.50 per million tokens for the whole request; Batch API 50% off ($0.05 / $0.25, or $0.25 / $1.25 above 100,000 tokens); US-only inference at 1.1x on all token categoriesSource

1 price change since tracking began.

InputOutput

$0$0.20$0.40$0.60$0.80Oct 13, 2026Nov 6, 2026
Daily observed price in the LLM price index in US dollars per 1M tokens. The index uses a market price from several sources, which can differ from the developer's direct price.

Free to use at Anthropic (Claude Free), more from €21.42 a month (Claude Pro). Through other providers from €6.83 a month (Poe Basic, with limited points). 31 plans in Germany.

Filter plans
PlanMonthly priceNote
Claude FreeAnthropicfreedeveloper's own plan
Claude ProAnthropic€21.42developer's own plan
Claude Team Standard-PlatzAnthropic€25.04developer's own plan, with limited credit, for teams, per seat
Claude MaxAnthropic€107.10developer's own plan, with limited credit
Claude Team Premium-PlatzAnthropic€125.22developer's own plan, with limited credit, for teams, per seat
Claude UltraAnthropicn/adeveloper's own plan, with limited credit
Kiro StudentsAmazon Web Servicesfreewith limited credit
Poe Basic€6.83with limited credit
Command Code GOAT≈ €10.60with limited credit
GitHub Copilot Pro≈ €10.60with limited credit
GitHub Copilot Business≈ €20.14with limited credit, for teams, per seat
Command Code Pro≈ €21.20with limited credit
Kiro ProAmazon Web Services≈ €21.20with limited credit, per seat
Replit Core≈ €21.20
Poe Plus€27.36with limited credit
Langdock Business Standard€29.75for teams, per seat
GitHub Copilot Enterprise≈ €41.35with limited credit, for enterprises, per seat
GitHub Copilot Pro+≈ €41.35with limited credit
Kiro Pro+Amazon Web Services≈ €42.41with limited credit, per seat
Poe Pro€68.41with limited credit
Command Code Max 10×≈ €106.02with limited credit
GitHub Copilot Max≈ €106.02with limited credit
Kiro Pro MaxAmazon Web Services≈ €106.02with limited credit, per seat
Replit Pro≈ €106.02
Langdock Business Max€117.81for teams, per seat
Poe Advanced€136.84with limited credit
Command Code Max 20×≈ €212.05with limited credit
Kiro PowerAmazon Web Services≈ €212.05with limited credit, per seat
Poe Max€342.11with limited credit
Cursor Enterprisen/afor enterprises
Kiro EnterpriseAmazon Web Servicesn/afor enterprises

All plans in the AI subscription comparison

Frequently asked questions about Claude Haiku 5.5

How good is Claude Haiku 5.5 in benchmarks?

Good: Rank 20 of 91 in the overall index.

Overall index (Gradually Meta-Score v4) from 45 benchmark results · data through Nov 6, 2026

IntelligenceOverall index in points, 100 = reference group average · higher is better
DeepSeek V4.1 Flash 1017106.6
Gemini 3.8 Flash106.5
GPT-6.1 Terra105.2
Claude Haiku 5.5Rank 20103.4
Grok 4.7103.3
Grok 4.6102.4
Kimi K3101.2

What is the Overall index (Gradually Meta-Score v4)?

A point value that combines many independent benchmarks into one number. 100 points is the average of a fixed reference group, higher is better. Claude Haiku 5.5 scores 103.4 points, which is rank 20 of 91.

Own analysis by gradually.ai, method benchmark-meta-score-v4 · for Claude Haiku 5.5 from 45 test series

Which data
Only current, reviewed results from independent evaluators such as Artificial Analysis, Vals AI, LMArena, LiveBench and Epoch AI. Vendor claims and historical or disputed values do not count.
One value per test
For each source and test series the latest retrieval with one fixed setup counts. When a source tests several reasoning levels, the highest documented one applies. A series needs at least 5 models.
Making tests comparable
Percentages are converted so that gaps near 0 and near 100 % count fairly, then every series is put on the same scale. A hard test therefore does not weigh more than an easy one.
Shared strength
A statistical model estimates, from all series at once, one shared capability per model and, per area, how far a model deviates from it. Missing tests are never counted as zero.
Weighting
Core areas (Coding, Reasoning, Mathematics, Knowledge, Agents, General and Multimodal) count 75 % together, the domain areas Finance, Legal and Health 25 %.
The point scale
The value is measured against a fixed reference group of 29 models, much like an IQ score: 100 points is its average and 15 points one standard deviation of the group. The scale is open at both ends and is not a percentage. A model 15 points ahead does better in roughly three of four benchmark comparisons. New models do not shift older values.
Place by score
The place follows the order of the scores among all models with a place, such as "rank 14 of 95". A second calculation checks whether a lead is reliable by recomputing the score 200 times with randomly redrawn test families. If a model is not almost always ahead there, the two models count as level.
Minimum evidence
A value needs at least 8 test series from 2 sources and 3 areas, a place only with enough precision. Missing core areas are flagged as "estimated".
What it is not
Not a test run by gradually.ai and no verdict on every single task. For a specific task the area values and the individual benchmarks matter more.

How much does Claude Haiku 5.5 cost?

Through the API $0.10 per 1M input tokens and $0.50 output. That is cheap.

Prices from Nov 6, 2026

Output price since tracking beganDollars per 1M tokens · lower is better

Oct 13, 2026Nov 6, 2026

First price on Oct 13, 2026: $0.55. Since Oct 14, 2026: $0.50.

What is the knowledge cutoff of Claude Haiku 5.5?

June 2026, according to Anthropic. That is 4 months before release.

Stated by Anthropic

How large is the context window of Claude Haiku 5.5?

1 M tokens. That is how much text Claude Haiku 5.5 reads at once, about 1,500 A4 pages.

Page count as a rule of thumb from Artificial Analysis

Context windowTokens the model reads at once · higher is better
GPT-6.1 Terra1.05M
Gemini 3.8 Flash1.05M
DeepSeek V4.1 Flash 10171.05M
Kimi K31.05M
Claude Haiku 5.5Rank 31 of 1011M
Grok 4.7500K
Grok 4.6500K

How fast is Claude Haiku 5.5?

236 tokens per second. That is faster than a typical model at 87.

Measured by Artificial Analysis on Nov 6, 2026. AA 3.0.0, including time to first token, averaged across available workloads and five days.

SpeedTokens per second · higher is better
Claude Haiku 5.5Rank 5 of 100236
DeepSeek V4.1 Flash 1017172
Gemini 3.8 Flash115
GPT-6.1 Terra73
Grok 4.672
Grok 4.771
Kimi K339

Can I run Claude Haiku 5.5 locally?

No. Anthropic does not publish the weights.

Stated by Anthropic

Open weights
no
License
n/a
Parameters
n/a
active per token
n/a

Where can I use Claude Haiku 5.5?

Free to use at Anthropic. Including other providers that makes 31 plans in Germany, plus the API.

Subscription comparison, prices in Germany incl. VAT

At Anthropic

6 plans, free (Claude Free).

Through other providers

25 plans from 7 providers, from €6.83 a month (Poe Basic), with limited credits.

Through the API

$0.10 input, $0.50 output per 1M tokens, unchanged since Oct 14, 2026.

Is Claude Haiku 5.5 better than Claude Haiku 4.5?

Yes, clearly in the overall index. It leads in 6 of 6 areas tested for both.

Claude Haiku 4.5 as predecessor according to Anthropic Claude Haiku 5.5 migration guide, checked Oct 13, 2026

IntelligenceOverall index, 100 points = average · higher is better
103.4

Claude Haiku 5.5

Rank 20

53.7

Claude Haiku 4.5

Rank 79

SpeedTokens per second · higher is better
236

Claude Haiku 5.5

fast

97

Claude Haiku 4.5

fairly fast

PriceDollars per 1M tokens, 3 parts input to 1 part output · lower is better
$0.20

Claude Haiku 5.5

cheaper

$2

Claude Haiku 4.5

10 times the price

Haiku 5.5Haiku 4.5100 points = average, higher is better
CodingHaiku 5.5 aheadHaiku 5.5: 106.2Haiku 4.5: 48.9
ReasoningHaiku 5.5 aheadHaiku 5.5: 105.2Haiku 4.5: 52.3
Mathematicsonly Haiku 5.5 has a scoreHaiku 5.5: 107.1Haiku 4.5: not enough data
KnowledgeHaiku 5.5 aheadHaiku 5.5: 85.5Haiku 4.5: 60.2
AgentsHaiku 5.5 aheadHaiku 5.5: 106.1Haiku 4.5: 60.1
GeneralHaiku 5.5 aheadHaiku 5.5: 106.4Haiku 4.5: 52.1
FinanceHaiku 5.5 aheadHaiku 5.5: 104.7Haiku 4.5: 58.0
Legalonly Haiku 5.5 has a scoreHaiku 5.5: 94.4Haiku 4.5: not enough data
Healthonly Haiku 5.5 has a scoreHaiku 5.5: 108.9Haiku 4.5: not enough data

Sources and data date

Show all 100 sources

Benchmark sources at a glance

Which independent evaluators tested this model. This is coverage, not a quality ranking.

EvaluatorTest seriesComposite indicesRecorded active measurements
Artificial Analysis13067
Vals AI285285
LMArena2029
LiveBench7146
Epoch AI6011
ARC Prize009
Scale AI Evals004

Without vendor claims and historical or disputed results; those stay labelled in the single values. 22 measurements are not yet assigned to a test series.

TypeEvidence and data date
Speed

Anthropic, Max effort. AA 3.0.0, including time to first token, averaged across available workloads and five days. Retrieved November 6, 2026.

Source: Artificial Analysis · Speed (Artificial Analysis)Speed measurement methodology
Price

Base API prices without caching or batch discounts. Higher context tiers and other rates are listed under Costs.

Source: Pricing · Model metadata (retrieved October 13, 2026) · Claude Haiku 5.5 model overview · Editorial description reviewed October 13, 2026
Input

Documented modalities. Retrieved October 13, 2026.

Source: Model documentation · Model metadata (retrieved October 13, 2026) · Claude Haiku 5.5 model overview · Editorial description reviewed October 13, 2026
Output

Documented modalities. Retrieved October 13, 2026.

Source: Model documentation · Model metadata (retrieved October 13, 2026) · Claude Haiku 5.5 model overview · Editorial description reviewed October 13, 2026
Research statusResearch date October 13, 2026. 54 source URLs checked. This documents the inspected sources, not an exhaustive inventory of every publication.
Unresolved

benchmark-results

Note

Exact configured results from the Claude Haiku 5.5 system card are recorded, including the separately identified SWE-bench variants. This source review did not establish exact GPQA or Vals AI results. Missing task-specific denominators, intervals and setup details remain identified on the individual observations.

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 published

numeric confidence intervals for graph-only results

Note

Some figures show 95% intervals but exact endpoints are not stated in searchable text; none are inferred from plotted bars.

Not published

unreported task denominators

Note

Only exact scored task or attempt counts stated in the card are retained. Language totals and broad corpus sizes are not substituted for model-specific N.

Additional sourceArtificial Analysis (retrieved October 19, 2026; November 2, 2026; November 6, 2026)12345678
Additional sourceArtificial Analysis (retrieved October 19, 2026; October 29, 2026; November 2, 2026; November 6, 2026)12
Additional sourceArtificial Analysis (retrieved October 19, 2026)12
Additional sourceArtificial Analysis (retrieved October 19, 2026; November 2, 2026)12
Additional sourceArtificial Analysis (retrieved November 5, 2026; November 6, 2026)
Additional sourceArtificial Analysis (retrieved November 6, 2026)12
Additional sourceLMArena (retrieved October 30, 2026)12
Additional sourceLMArena (retrieved October 28, 2026)12
Additional sourceVals AI (retrieved November 6, 2026)123456789101112131415161718192021222324252627
Additional sourceVals AI (retrieved October 16, 2026; November 6, 2026)1234
Additional sourceVals AI (retrieved October 16, 2026; October 31, 2026; November 6, 2026)
Additional sourceVals AI (retrieved October 28, 2026; November 6, 2026)
Additional sourceLiveBench (retrieved October 16, 2026)1 (unavailable on November 2, 2026) https://livebench.ai/table_2026_09_02.csv2 (unavailable on November 2, 2026) https://livebench.ai/breakdown_2026_09_02.json
Additional sourceAnthropic Claude Haiku 5.5 System Card (retrieved October 17, 2026)
Additional sourceVulcanBench SWE-v4 evaluator report, Code Quality v3.24 (retrieved October 17, 2026)
Additional sourceAnthropic, Claude Haiku 5.5 product page (AlphaSense customer evaluation) (retrieved October 17, 2026) · Anthropic, Introducing Claude Haiku 5.5 · Editorial description reviewed October 13, 2026
Additional sourceDuelLab GameBench 2 model results (retrieved October 17, 2026; October 28, 2026; November 2, 2026; November 5, 2026)
Additional sourcePlotly Studio Embedded model benchmark, Claude Haiku 5.5 (retrieved October 17, 2026)
Additional sourceARC Prize official results, Anthropic Claude Haiku 5.5 (retrieved October 17, 2026)
Additional sourceSolving VeriContest with a Lean-Backed Rust Verifier (retrieved October 18, 2026)
Additional sourceEpoch AI benchmark run dataset snapshot (retrieved October 19, 2026; October 28, 2026)
Additional sourceRoboflow Vision Evals (retrieved October 19, 2026)
Additional sourceAnthropic Transparency Hub (retrieved October 19, 2026)
Additional sourceLiveBench (retrieved October 28, 2026)12
Additional sourceParallel Search Capability Leaderboard (retrieved October 28, 2026; November 5, 2026)
Additional sourceTowards AI ToneBench (retrieved October 28, 2026; October 30, 2026)12345
Additional sourceRoboflow Vision Evals (retrieved October 28, 2026)
Additional sourceAnthropic Claude Haiku 5.5 System Card (retrieved October 28, 2026)
Additional sourceGoogle on Kaggle (retrieved October 28, 2026)123
Additional sourceGoogle DeepMind on Kaggle (retrieved October 28, 2026)
Additional sourceMathBench.ai (retrieved October 28, 2026)
Additional sourceScale AI (retrieved October 28, 2026)
Additional sourceScale AI MultiChallenge (retrieved October 28, 2026)
Additional sourceScale AI FORTRESS (retrieved October 28, 2026)
Additional sourceScale SWE-Bench Pro V2 public leaderboard (retrieved October 28, 2026)
Additional sourceParallel Search Capability Leaderboard (retrieved November 5, 2026)
Additional sourceModel metadata (retrieved October 13, 2026)
Additional sourceAnthropic supported watermarking models (retrieved November 5, 2026)
Additional sourceAnthropic text watermarking method (retrieved November 5, 2026)
Additional sourceWhat's new in Claude Haiku 5.5 · Editorial description reviewed October 13, 2026
Additional sourceOpenRouter: Popularity index (derived), CC BY 4.0
Additional sourceVercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional sourceOllama: cumulative downloads for the popularity index
Additional sourceReplicate: cumulative runs for the popularity index
Additional sourceGoogle Trending Now: limited feed selection for Germany and the US
Additional sourceCreative Commons Attribution 4.0
Additional sourceAnthropic Claude Haiku 5.5 migration guide (2026-10-13)