Anthropic launches Claude Haiku 5.5, cuts small-model API prices ~75%
The new small model adds adjustable effort settings and posts large benchmark gains, while Anthropic also trims Sonnet 5.5 pricing.
Part of a larger narrative
Anthropic's Claude Evolution
- Anthropic builds Scion, a git replacement, as Claude pushes merges to 30K per day
- Anthropic warns small AI models could self-replicate onto smart home devices by 2026 end
- Anthropic discloses Claude AI sent a fake murder tip to Philadelphia police
- Anthropic cuts Claude reasoning-token theft by up to 99%, flags Chinese rivals
- Anthropic launches Claude Haiku 5.5, cuts small-model API prices ~75%you are here
What to know
- Haiku 5.5 runs about 75% cheaper than Haiku 4.5 and is Anthropic's first Haiku-class model with an adjustable effort setting.
- Anthropic also halved Sonnet 5.5 cache-read prices (~20% cheaper agentic work) and added monthly API credits for Claude Max and Team subscribers.
- Benchmark results show large jumps over Haiku 4.5 (e.g., OSWorld 15.7%→72.4%, HLE 10.2%→45.9%), with commentators noting it approaches Sonnet 5.5-level performance at its highest effort setting.
- Coverage frames the pricing as converging with OpenAI's GPT-6 Luna, with at least one outlet disputing whether Haiku 5.5 undercuts or merely matches Luna's price.
“At HubSpot, we use simulated portals to evaluate new models on CRM tasks like reporting on deals. ... Claude Haiku 5.5 got the best score we've seen on this suite yet, at 92.8% averaged over three runs.”
HubSpot, customer testimonial · Anthropic blog post ↗ · Oct 6
Anthropic AI company releasing the modelHubSpot customer cited in launch materialsBox customer cited in launch materials (Box AI)@chrisgpt independent commentator on X@viticci independent commentator (Federico Viticci) on X
How it unfolded 4 developments, newest first · click a bar or a number to jump articlespostscomments
-
3
@viticci notes Haiku 5.5 Max nears Sonnet 5.5 performance
Federico Viticci posts that Haiku 5.5 run at its highest effort setting approaches Sonnet 5.5's benchmark performance, sometimes at lower cost.
“Haiku 5.5 Max reaching Sonnet 5.5 High performance in these benchmarks - sometimes with lower costs. Fascinating.”
— @viticci -
Ran our DataAnalyticsBench benchmark on it: https://plotly.com/blog/claude-haiku-5-5-plotly-data-analyti...9x cheaper than Haiku 4.5 and 2 letter grades better. It's also now the fastest model (using the default speeds, not trying any of the other models "Fast" mode) to complete the exam.Similar ballpark to Luna in price, cost, and accuracy. These…
2 more of the top 3 · 46 posts in this stretch
-
Anthropic has released Claude Haiku 5.5, scoring 43 on the Artificial Analysis Intelligence Index - up 26 points one year after the last Haiku release Haiku 5.5 is the first Haiku model with Anthropic's effort settings and adaptive thinking, and Anthropic has introduced tiered pricing. Haiku 5.5 ...
-
Claude Haiku 5.5: Anthropic cuts API prices by up to 90 percent https://www. heise.de/en/news/Claude-Haiku- 5-5-Anthropic-cuts-API-prices-by-up-to-90-percent-11480285.html?utm_source=flipboard&utm_medium=activitypub Posted into heise online international @ heise-online-international-heiseonline
-
-
background
@chrisgpt highlights 3-5x benchmark jumps over Haiku 4.5 — An independent analysis posted on X notes Haiku 5.5's scores rising sharply versus Haiku 4.5 across several benchmarks, including OSWorld and Humanity's Last Exam, and claims it beats GPT-6 Luna on some measures.
-
4
VentureBeat ties Haiku 5.5 pricing to GPT-6 Luna
Coverage frames the new price cut as bringing Haiku 5.5's API pricing in line with OpenAI's GPT-6 Luna, with one outlet instead arguing the price matches Luna's rather than undercutting it.
-
6 outlets Anthropic's New Claude Haiku 5.5 Cuts AI Token Prices Up to 90%; Beats OpenAI's GPT-6 Luna
first by WinBuzzer, 6d ago · also Neowin, The Decoder, Simon Willison, Simon Willison's Weblog, TechRepublic
5 more headlines
- Anthropic launches Claude Haiku 5.5 with aggressive pricing, but it consumes far more tokens Neowin · 6d ago
- Anthropic's Claude Haiku 5.5 Matches GPT-6 Luna at 10 Cents per Million Input Tokens Implicator.ai · 6d ago
- Claude Haiku 5.5 arrives with massive price cuts proving the AI pricing arms race is far from over The Decoder · 6d ago
- Introducing Claude Haiku 5.5 Simon Willison · 6d ago
- Anthropic Launches Claude Haiku 5.5 With API Prices Starting at $0.10 TechRepublic · 4d ago
-
first by iClarified, 6d ago · also TestingCatalog AI News, VentureBeat
2 more headlines
- Anthropic launches Claude Haiku 5.5 at lower API prices TestingCatalog AI News · 6d ago
- Anthropic launches Claude Haiku 5.5 with 90% API price reduction, matching GPT-6 Luna VentureBeat · 6d ago
-
Haiku 5.5 is noticeably smarter than GPT-6 Luna, so I can see their pricing strategy here.For a while Anthropic has lacked a cost effective “cheap” LLM for summarisation, compacting, RAG helpers, etc.These ‘ephemeral’ workloads are often under 100k tokens, or can be structured to be under 100k.In some coding benchmarks, Haiku 5.5 beats Sonnet 5!…
2 more of the top 3 · 5 posts in this stretch
-
> Second, this week, we’ll roll out a new monthly API credit to all Max and Team subscribers for use on the Claude Platform. Max 5x users will get $100 in credits per month, Max 20x users will get $200, and Team subscribers will receive up to $500, pooled across their usersThis is a very big benefit for me. I can now ship actual ai enhanced…
-
There's also a tokenizer efficiency difference: modern Claude's 100K tokens are about ~60-65K modern GPT tokens, so in reality the Luna cutoff is much further away than the Haiku one.You can test with Anthropic's count_tokens endpoint or with
-
-
2
Anthropic launches Claude Haiku 5.5
Anthropic unveils Haiku 5.5 as its cheapest, fastest and most capable small model yet, built for high-volume tasks like summaries, classification, extraction and subagent work, running about 75% cheaper than Haiku 4.5. It also halves Sonnet 5.5 cache-read pricing (~20% cheaper agentic work overall) and adds a new monthly API credit for Claude Max and Team subscribers.
“Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we've ever released.”
— Anthropic -
8 outlets first by Decrypt, 6d ago · also 9to5Mac, XDA Developers, Yahoo Finance, Moneycontrol, The Tech Portal, Android Authority +1 · read ↗
-
4 outlets first by RuntimeWire, 6d ago · also Anthropic, Seeking Alpha, Investing.com News · read ↗
-
Claude Haiku 5.5, our latest model:
2 more of the top 3 · 4 posts in this stretch
-
Pricing is...a bit weird. Input $0.10 / MTok for prompts up to 100,000 tokens $0.50 / MTok for prompts over 100,000 tokens Output $0.50 / MTok for prompts up to 100,000 tokens $2.50 / MTok for prompts over 100,000 tokens 100k tokens is an absurdly low cutoff and it is only applicable to Haiku and not Sonnet or Opus. It's a low enough cutoff that…
-
Introducing Claude Haiku 5.5: the cheapest, fastest, and most capable small model we’ve ever released. On average, it costs around 75% less to run than Claude Haiku 4.5.
-
-
1
Anthropic publishes Haiku 5.5 technical documentation
Docs detail a 1M-token context window, up to 128k output tokens, an adjustable effort parameter, and a newer tokenizer shared with Claude 4.7-and-later models that counts roughly 30% more tokens than Haiku 4.5 for the same text.
What people are saying 18 voices from 3 sites · best of 55 · verbatim
- Oct 8
-
Very important caveat if looking at the Artificial Analysis cost vs. intelligence graph (https://artificialanalysis.ai/?models=gpt-6-luna-low%2Cclaud...):"The site does not yet reflect tiered pricing, so provisional cost figures for Haiku 5.5 do not include the step up cost."Frankly, it's irresponsible for them to put the points on the chart if…
- Oct 7
-
Some attempts to benchmark and compare LLM's:https://artificialanalysis.ai/https://benchlm.ai/https://tokenscost.comhttps://www.vellum.ai/llm-leaderboardAre there any projects out there for doing something similar?
-
thinking_effort: max Reasoning trace: This is the classic pelican-on-bicycle SVG test.Opus 5.5 had similar response on max: This is a classic test requesthttps://tools.simonwillison.net/markdown-svg-renderer?url=ht...I think your test is already embedded into the models. You should search for new frontier tests to subject the models to. Maybe they…
-
It's around Qwen-3.8, and Sonnet 5.5 level, but a lot cheaper. It is also really fast.My tests for Haiku 5.5: https://aibenchy.com/compare/anthropic-claude-haiku-5-5-xhig...Twice as expensive as Luna, but also considerably smarter too:
-
N
Claude Haiku 5.5: https://www. anthropic.com/claude-haiku-5-5 Discussion: http:// news.ycombinator.com/item?id=4 9996437
-
My weekly limit __on a Pro sub__ has not gone over 50% since before the pre-Fable promos, but usage has been pretty much the same from my point of view. Maybe I am holding it right? Anyone else getting this?As such, I do not need to even reach for Haiku, and 4.5 was so inaccurate that it often cost more to do so in the past. Sonnet 5.5/low has…
-
Pelicans riding bicycles for Haiku at the different thinking levels: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...Low messes up the bicycle frame, but medium/high/xhigh/max all get the bicycle frame right.The max one took 5 minutes 9 seconds and cost 3.3826 cents. The cheapest one (low) cost 0.0936 cents and took 7 seconds.The…
-
My complaint about Haiku 4.5 was that it was 10x the price of GPT-6 Luna.> Claude Haiku 5.5 is priced 90% lower than Claude Haiku 4.5 for requests up to 100,000 tokens, and 50% lower for requests over 100,000 tokensHaiku and Luna now have the exact same price up to 100,000 tokens. Luna is now cheaper for anything after 100,000 tokens, even after…
-
Anthropic announced haiku 5.5, I go into Claude code, it's already there OpenAI announced gpt 6 in chat, I check and it's still 5.6
-
Ran image -> html tests for this. I was curious if this smaller model was good enough for complex UI. It was not.Haiku 5.5: https://html.non.io/lcars-haiku-5.5/Opus 5.5 for comparison: https://html.non.io/lcars-opus-5.5Designs it was building from: https://diffui.ai/app/canvas/5093e689-1e74-4f26-b632-2a4500f...One interesting thing is it took a…
-
About time Anthropic released a competitive cheap model. Haiku 4.5 has been too expensive compared to its performance for months now (in fact I don't remember being too impressed even when it was released). This one actually looks worth using in some scenarios. If it's really as much of a step up from Luna as the benchmarks they've shown indicate…
-
Haiku 5.5 is an almost exponential jump over the last Haiku generation, with several benchmarks improving by 3-5x over Haiku 4.5 Haiku 5.5 goes from 15.7% -> 72.4% on OSWorld, 10.2% -> 45.9% on HLE without tools, 18.7% -> 57.4% with tools, and 6.4% -> 46.4% on Chartography. It even beats GPT-6 L...
-
Anthropic just launched Claude Haiku 5.5, with average running costs around 75% below Haiku 4.5, according to the company. 5.5 Way better than GPT-6-Luna, close(r) to Sonnet 5.5 Anthropic says it also improves coding, computer use, and knowledge work. Haiku now has adjustable effort, letting de...
-
Haiku 5.5 makes absolutely no sense lol. Anthropic somehow made Haiku MUCH smarter... then made it ~75% cheaper to run. and it's borderline better than Opus 5: Haiku 5.5 vs Opus 5: > GDPval: 1620 vs 1596 > AA-Briefcase: 1578 vs 1562 > HLE: 45.9% vs 53% > Terminal-Bench 4.0: 39.2% vs 46% yes....
-
Claude Haiku 5.5 appears to be a massive jump in both agentic, coding & reasoning capability & incredible cost efficiency. In fact so much so that it is now competitive in cost with & even better than, similarly capable open-source models. It's crazy how Anthropic made that leap!
-
Haiku 4.5 came out two months after I joined Anthropic. It scored 15.7% on the test for operating a computer. Haiku 5.5 is out today, less than a year on. It scores 72.4% and costs around 75% less to run.
-
Claude Haiku 5.5 is out 🎉 It's a cheap, fast and very capable small model. ~75% cheaper to run than Haiku 4.5 with an adjustable effort setting. My team loves it as a subagent alongside Opus 5.5.
-
Anthropic just dropped Haiku 5.5. Their fastest Claude ever. 90% cheaper than Haiku 4.5 for most requests. AI agents just got way cheaper.