Just now, Claude Haiku 5.5 was released! The price plummeted by 90%

Just now, Anthropic released Claude Haiku 5.5, priced at only 1/10 of the previous generation Haiku 4.5, 1/20 of Sonnet 5.5, and the same price as OpenAI's GPT-6 Luna.

In the official benchmark table, Luna lost all six categories to it, with several categories closely trailing Sonnet.

The OSWorld for operating computers jumped from 15.7% in the previous generation to 72.4%, just 11.5 percentage points away from Sonnet 5.5.

The Terminal-Bench 4.0 for terminal programming scored 39.2%, up from zero in the previous generation.

The GDPval-AA for knowledge work scored 1620 points, more than double that of the previous generation.

The "Final Exam for Humanity" HLE, without tools, rose from 10.2% to 45.9%.

How Much Can You Save

The price is $0.10 per million tokens for input and $0.50 for output, which is the price for prompts under 100,000 tokens; for over 100,000, it is calculated at $0.50 for input and $2.50 for output.

Cache reading costs $0.01 per million tokens, and batch processing gets an additional 50% discount.

According to official calculations, the same tasks average about 75% cheaper than Haiku 4.5, allowing approximately four times the call volume for the same budget.

The savings are not 90% because requests over 100,000 tokens are only half the price, but fortunately, about 90% of requests in the previous generation Haiku were under 100,000 tokens. Additionally, the new tokenizer will increase the number of tokens for the same passage by about 30%.

Compared to GPT-6 Luna, the input, output, and cache prices for both are exactly the same for under 100,000 tokens.

When dealing with long contexts, Luna is cheaper; it only raises prices around 270,000 tokens, and even then, it charges $0.20 for input and $0.75 for output.

Sonnet 5.5's cache reading has also been halved starting today, dropping from $0.20 per million tokens to $0.10, with the official calculation showing that most Agent tasks can be about 20% cheaper.

Max and Team subscriptions will start offering API credits monthly from this week, with Max 5x providing $100, Max 20x providing $200, and Team up to $500, shared among the whole team, usable on any model on the Claude Platform.

Scored Higher than Luna but Still Lags Behind Sonnet in Complex Programming

This generation of Haiku has made significant progress compared to the previous generation.

The OSWorld for operating computers jumped from 15.7% in the previous generation to 72.4%, just 11.5 percentage points away from Sonnet 5.5.

The Terminal-Bench 4.0 for terminal programming scored 39.2%, up from zero in the previous generation.

The GDPval-AA for knowledge work scored 1620 points, more than double that of the previous generation.

The "Final Exam for Humanity" HLE, without tools, rose from 10.2% to 45.9%.

Computer operation is the area with the most significant improvement this time, with Luna at the same price only scoring 48.9%, while Haiku 5.5 outperformed it by 23.5 percentage points.

When considering costs, the difference is even more dramatic; it exceeded Luna's maximum score when set to Medium mode, while costing less.

In knowledge work, Haiku 5.5's score of 1620 surpassed Luna's 1437, still trailing Sonnet 5.5's 1840 by a bit, but even at the default Medium setting, it scored 1277.

In long-term project scoring on AA-Briefcase, it rose from 614 points in the previous generation to 1578 points, while Luna scored 1336.

In reasoning and image reading, when HLE uses search and code tools, Haiku 5.5 scored 57.4%, while Sonnet 5.5 scored 64.5%. In reading professional charts, Chartography rose from 6.4% in the previous generation to 46.4%, while Luna scored 29.1% and Sonnet 5.5 scored 61.6%.

The biggest gap with Sonnet is in programming; on Terminal-Bench 4.0, its 39.2% is more than double Luna's (16.4%), but Sonnet 5.5 scored 70.6%.

On FrontierCode, it scored 46.4%, beating Luna's 42.4%, but still nearly 6 points behind Sonnet 5.5's best score of 52.1%.

The official also stated that complex Agent programming is still best left to Sonnet 5.5 and Opus 5.5.

What It Is Suitable For

The official positioning for Haiku 5.5 is for large-scale, fast, and cost-effective tasks, such as writing summaries, compressing contexts, querying databases, and classification.

It is also currently the fastest Claude model, only slightly slower than Opus when set to fast mode (which is very expensive!), making it suitable for real-time customer service and browser operations.

The official also provided feedback from a batch of star AI application companies during internal testing.

Asana found that task completion delays were reduced by over 30% compared to their current model, while Box measured its delay to be about half that of Haiku 4.5.

Another main use is as a sub-agent for Opus or Sonnet, where the large model breaks down tasks and makes decisions, while Haiku runs them.

Cognition's Devin Fusion uses Opus 5.5 as the main model and Haiku 5.5 as the assistant, maintaining a top-level score of 66.2, with reduced costs and delays.

The financial AI company Rogo uses the large model to create PPTs, with Haiku 5.5 diving into 10-K reports to extract revenue data by segment.

The Claude developer account provided a demo, with the task of designing an egg holder using a bunch of household items, aiming to drop it from 32 meters without breaking.

Opus 5.5 completed it in 3 minutes and 37 seconds, trying 25 solutions and costing $0.47, while with 10 Haiku 5.5 sub-agents, it only took 58 seconds, tried 86 solutions, and cost $0.14.

Haiku 5.5 User Guide

Haiku 5.5 is now available on the Claude client, Claude Code, and platforms like AWS, Google Cloud, Microsoft Azure, Cursor, OpenRouter, etc. The model ID is claude-haiku-5-5, with a context of 1 million tokens and a maximum output of 128,000 tokens per request.

It is the first Haiku that can adjust thinking levels, with five levels from Low to Max, and the API defaults to Medium.

Python and TypeScript SDKs have also added computer operation and browser operation, currently in Beta.

When migrating from Haiku 4.5, the same text tokens will increase by about 30%, and max_tokens and cost estimates need to be recalculated.

budgettokens have been replaced with adaptive thinking and Effort levels, while temperature, topp, and top_k need to be removed, and pre-filled responses are no longer usable; format requirements have changed to structured output.

Computer operations need to switch to the new toolset computertoolset20260801, and Priority Tier is not yet supported.

If you find it troublesome, you can run a command in Claude Code:

Let it modify itself.

Of course, for non-developers, you can now directly converse with Haiku 5.5 in the client!

For simple questions, there's no need to bother Fable, Opus, or Sonnet. Haiku 5.5 can provide accurate answers even with low inference intensity, perfectly replacing search engines.

Go try it out!

References:

https://www.anthropic.com/claude-haiku-5-5