Advertise here
#120

Claude Haiku 5.5

Anthropic Release: 2026-10-07 Tested on: 2026-10-07 21:50 anthropic/claude-haiku-5.5::medium
~50BDenseClosedEstimated
(max) (xhigh) (high) (medium) (low)

Summary

Claude Haiku 5.5 scores 7.7 on AI BENCHY and ranks #120. It has 10.0 reliability, a 71.0% pass rate, $0.063 total cost, and 7.71s average response time.

What makes Claude Haiku 5.5 unique: It stands out most in Instructions following, where it ranks #1, while General Intelligence is its weakest area at #17. It is notably fast compared with similar models.

Model facts

Researched on 2026-10-07

Estimated
Parameters
~50B
Architecture
Dense
Availability
Closed
License
-

Best estimate from public evidence; the vendor did not disclose every value. Anthropic does not disclose parameter counts or architecture. The 50B dense entry is a low-confidence catalog estimate for the small, latency-oriented Haiku tier, not a vendor claim or a measured count. The cited announcement establishes the exact model identity, positioning and closed API availability. OpenRouter declares low, medium, high, xhigh and max effort levels with medium default and adaptive thinking.

Parameter source 1 Parameter source 2 Availability source

Score

7.7

Consistency

7.5

Reliability

10.0

Total Cost (Current Price)

$0.063

Total Output Tokens

66,174

Total Input Tokens

298,334

Input Price

$0.100 / 1M

Output Price

$0.500 / 1M

Cache Read Price

$0.010 / 1M

Cache Write Price

$0.125 / 1M

Cache prices apply to input tokens. Reads reuse cached prompts; writes store them and can cost extra. Output tokens use the output price.

Tests Correct

Wrong Tests: 10

Attempt pass rate: 71.0%

Flaky tests

7

Flaky tests had mixed outcomes across runs (at least one pass and one fail).

Response Time (avg)

7.71s

Response Time (max): 47.70s

Response Time (total): 177.29s

Wrong answer: 5 Extra formatting: 2 Did not follow instructions: 2 No answer: 1

Hamster playing table tennis

Prompt: Create a detailed SVG illustration of a hamster playing table tennis.

#120 Claude Haiku 5.5

medium
Cost
$0.002
Time
17.2s
Tokens
3,524 tok

Charts

Choose the first model, then click a second model to open a side-by-side page.

Total Output Tokens

Score vs Total Output Tokens

Quick Compare

Claude Haiku 5.5mediumvsMuse Glimmer 30BhighClaude Haiku 5.5mediumvsQwen3.5 Plus 2026-04-20mediumClaude Haiku 5.5mediumvsGPT-6 LunahighClaude Haiku 5.5mediumvsQwen3.7 FlashlowClaude Haiku 5.5mediumvsDeepSeek V4 Flash 0731mediumClaude Haiku 5.5mediumvsMuse Glimmer 30BmediumClaude Haiku 5.5mediumvsGrok 4.5lowClaude Haiku 5.5mediumvsQwen3.8 Flash NextmediumClaude Haiku 5.5mediumvsQwen3.7 PlusnoneClaude Haiku 5.5mediumvsGPT-6 Lunalow

Category Breakdown

Category Score Consistency Tests Correct
Agentic 10.0 10.0
Anti-AI Tricks 8.7 7.9
Coding 6.3 4.8
Combined 7.3 5.8
Data parsing and extraction 10.0 10.0
Domain specific 2.9 4.4
General Intelligence 3.8 2.5
Instructions following 10.0 10.0
Puzzle Solving 7.7 10.0
Tool Calling 10.0 10.0
Trivia 3.0 10.0

Compared models