None
None
Long Story Video Skill
Turn your idea into 10s–10min video—AI writes scripts, prompts, and generates footage automatically.
Ads Video Skill
Generate professional ads and sales videos—AI auto-generates scripts, prompts, and footage.
3D Science Explainer Video Skill
Convert scientific concepts into stunning 3D explain animations
Feedback
freeTrialImage.bannerPity
freeTrialImage.upgradeUnlock
- ✓freeTrialImage.benefitHd
- ✓freeTrialImage.benefitWatermark
- ✓freeTrialImage.benefitUnlimited
Haiku 5.5 Benchmark
An independent walk-through of Claude Haiku 5.5: published launch scores, token price tiers and hard limits, each figure tied to its config and pull date.
All Tools
Discover our comprehensive AI-powered animation toolkit
MiniMax H3
MiniMax H3 AI Video Generator
Seedance 2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
FLUX 3 Video Generator
Kling 3.0
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
MiniMax H3 video generator
MiniMax H3 AI Video Generator
Seedance 2.0
The Future of AI Video Is Here.
What Each Haiku 5.5 Benchmark Figure Is Really Saying
Launch results, token price tiers and per-effort costs, every one stored next to the configuration it ran on and the date it was pulled.
- Each Score Uses Its Own YardstickElo ratings, partial credit, coding pass rates and accuracy scores sit on scales that do not convert into one another, so averaging them means nothing.
- Built for Volume, Not One-Off PromptsExtraction, tagging and summarization jobs that run thousands of times a day are where the per-token math actually pays off.
- Two Footnotes That Rewrite the TablesThe low price tier ends at 100K prompt tokens, and the headline 72.4% computer-use figure counts partial credit rather than finished tasks.
Turning Haiku 5.5 Benchmark Data Into a Decision
Three steps that convert published figures into an answer for your own pipeline, not someone else's.
What the Haiku 5.5 Benchmark Actually Shows
Specs, token pricing tiers, coding outcomes, computer-use results and third-party index numbers for Claude Haiku 5.5, all in one pass.
Context and Output Ceilings
Text and images in, text out, with adaptive thinking enabled. A one-million-token window and a 128,000-token output ceiling make Haiku 4.5's 200,000 and 64,000 look modest.
Pricing Splits at 100K Prompt Tokens
Go past 100,000 prompt tokens and input, output and caching all shift into a tier five times higher. By contrast, OpenAI's standard band reaches 272,000 input tokens.
Terminal-Bench 4.0: Finished Tasks
The 39.2% launch score clears Luna comfortably yet trails the published Sol entries. Each run allows an eight-hour agent timeout, and any 2.1 or 3.0 figure comes from an earlier experiment.
FrontierCode: Merge Quality vs. Effort Spend
Scoring weighs correctness, tests, scope, style and repo conventions. Max effort yields 46.4% and medium 41.6% at $0.13 per rollout, while climbing from xhigh to max doubles cost for half a point.
OSWorld 2.1: Credit vs. Completion
The suite is 82 offline tasks, five attempts each, 1080p screenshots and a 500-step cap. If a workflow has to actually finish, the 37.1% strict figure is the one to quote.
Third-Party Index and Tool Gains
Artificial Analysis lists all five effort levels on Intelligence Index v4.3.2 with rounded per-task costs. HLE moves from 45.9% unaided to 57.4% with tools, an uplift produced by the harness rather than the model alone.
Haiku 5.5 Benchmark Questions, Answered
Straight answers on Claude Haiku 5.5 scores, token pricing, context limits and the comparisons people keep asking about.
Where does Haiku 5.5 sit next to GPT-6 Luna and Sol?
On knowledge tasks, screenshot work and small coding jobs, Haiku and Luna trade wins. Sol and Sol 6.1 charge more and justify it on coding, while Sonnet 5.5 still tops most rows.
Why isn't the 72.4% computer-use figure a success rate?
Require every checkpoint to pass and the same run drops to 37.1%. A model can find and edit the right file, then save it in the wrong format or quit one action early. Always add an end-state assertion.
At what point does Haiku 5.5 move to the higher price tier?
Prompts up to 100,000 tokens qualify for the lowest rates; beyond that, input, output and cache all cost five times as much. The 90%, 50% and roughly 75% savings claims each measure against a different baseline.
Could Haiku 5.5 run locally, and what GPU memory would that take?
The published specs list no parameter count, no active-parameter figure, no layer count and no training compute. 'Small' describes a product tier rather than an architecture, so any local VRAM estimate is guesswork.
Does raising reasoning effort always pay off?
No. Sol 6.1 actually scores higher at medium than at max, and Haiku's 0.2-point lead over Sonnet at max sits within noise. More effort also means more tokens, more latency and more room to wander off task.
Why do identical models post different numbers across tables?
Sonnet appears at 70.6% in Anthropic's launch table but 61.8% on the public leaderboard snapshot, which carried no Haiku row at all. The 'Default Fallback' tag was later removed from Artificial Analysis, while Anthropic's migration guide continues to state that no server-side fallback exists.
Run the Haiku 5.5 Benchmark on Your Own Workload
Choose a single task, fix the effort setting, and see whether the numbers hold up against the bar your team already ships at.