Загрузка видео...

Не удалось загрузить видео

На главную

Haiku 4.5 is a workhorse that makes the coding experience in Claude Code feel really fast. While Sonnet 4.5 remains the default, Haiku 4.5 now powers the Explore subagent which can rapidly gather context on your codebase to build apps even faster.

205,593 просмотров • 1 год назад •via X (Twitter)

Комментарии: 40

Фото профиля cat
cat1 год назад

You can select Haiku 4.5 to be your default model in /model. When selected, you’ll automatically use Sonnet 4.5 in Plan mode and Haiku 4.5 for execution for smarter plans and faster results. To enter Plan mode, hit `Shift + Tab + Tab`

Фото профиля cat
cat1 год назад

Haiku 4.5 is a great way to extend your Claude Code usage on the Pro plan. Haiku 4.5 is $1 per million input tokens and $5 per million output tokens, which means it is priced 3x lower than Sonnet 4.5 and slightly higher than Haiku 3.5.

Фото профиля cat
cat1 год назад

Let us know how it feels to use Haiku 4.5 in Claude Code. We’re excited to keep making it better.

Фото профиля J Liew
J Liew1 год назад

Claude Code v2.0.17 I don't see the option to choose Haiku 4.5 using /model

Фото профиля Alex Finn
Alex Finn1 год назад

amazing! needed a lightweight Anthropic model for the agent I'm building. Playing around with it all day

Фото профиля cat
cat1 год назад

let us know what you think!

Фото профиля elvis
elvis1 год назад

Cool! It does sound like Haiku 4.5 will be great to build subagents. Can't wait to test this out with my subagents.

Фото профиля Eng1neerJack
Eng1neerJack1 год назад

@claudeai is making people hit pay barriers more and more - you have to keep paying more and more to get services. This might be good for now - but soon im sure another company will replace it. If claude keeping being tight - it will be replaced.

Фото профиля CJ
CJ1 год назад

We don’t need new models. Fix the ones you nerfed instead.

Фото профиля Mukul. 🚀
Mukul. 🚀1 год назад

How is Haiku on writing tasks? If I wish to repurpose CC to be more of a researcher or writer agent, will this be a good segue?

Фото профиля cat
cat1 год назад

Yes, I use CC for this :)

Фото профиля Mukul. 🚀
Mukul. 🚀1 год назад

Exciting! Thanks for building a great product :)

Фото профиля Aditya Saxena
Aditya Saxena1 год назад

i'm on the max plan but my usage limits don't reset anymore (been seeing this error since 2 days):

Фото профиля thoughtlesslabs
thoughtlesslabs1 год назад

welp - there goes my week

Фото профиля Conwic
Conwic1 год назад

Does it contain lead?

Фото профиля Vishnu Anand
Vishnu Anand1 год назад

But actually it is indeed a good model to try building agents…

Фото профиля Alexander Riccio (@co2trackers)
Alexander Riccio (@co2trackers)1 год назад

Quick question: does haiku 4.5 have the same 200k token context window limit?

Фото профиля Jon Chu
Jon Chu1 год назад

For bug fixes seems like haiku will be the way to go

Фото профиля Cody Hergenroeder 🍋⚡️
Cody Hergenroeder 🍋⚡️1 год назад

YOOO this is so handy to have in the subagent repertoire for extending claude's capabilities :') gonna use haiku in much the same way you are, to extend claude's Fog of War view in massive knowledgebases!

Фото профиля Agam Chaudhary
Agam Chaudhary1 год назад

A strong sign of how AI tooling is maturing from capability to usability. Speed and context awareness are becoming the real differentiators. Do you think this kind of embedded agent design will redefine how developers approach problem solving?

Фото профиля Paul
Paul1 год назад

solid!!

Фото профиля Der.dev 🔥🛠️
Der.dev 🔥🛠️1 год назад

🫶🏿🧠

Фото профиля Benita
Benita1 год назад

I really didn’t think Speed would be such a game changer for me, that it would always be about using the “best” (smartest) model. I was wrong.

Фото профиля Lane
Lane1 год назад

Shame Claude doesn't actually work.. the few times it did work, worked well. Back to GPT.

Фото профиля Conor Dart
Conor Dart1 год назад

im trying to understand where Haiku 4.5 fits in claude code? can anyone help me understand better

Фото профиля BB🍓🍓🍓
BB🍓🍓🍓1 год назад

The latest version of Claude Code doesn’t have the Opus Plan mode anymore — is there any way to bring it back?

Фото профиля Casey Whalen
Casey Whalen1 год назад

Extremely excited to test out this new model.

Фото профиля Suhrab Khan⚡️
Suhrab Khan⚡️1 год назад

Haiku 4.5’s speed advantage and context-gathering capabilities make it ideal for rapid app development, complementing Sonnet 4.5 for deep code exploration.

Фото профиля Hayden Lingle
Hayden Lingle1 год назад

Too many limits. Had to get codex since it’s been so bad.

Фото профиля AICarnal
AICarnal1 год назад

I suspect that this model was used in the recent beta testing of Claude OS. I think it's ideal for web searches, and it's quite pleasant to converse with. Speed is very important, part of the future lies with Haiku.

Фото профиля Mehreen Omer
Mehreen Omer1 год назад

Simply beautiful

Фото профиля Caden Burleson
Caden Burleson1 год назад

Now edit a dropdown menu…

Фото профиля VibeEdge
VibeEdge1 год назад

Impressive speed and efficiency with Haiku 4.5! The demo really showcases its potential for rapid prototyping. Excited to see how it performs in real-world projects great work!

Фото профиля 🙃 Ben WΞst 🦋
🙃 Ben WΞst 🦋1 год назад

impressive, looking forward to seeing the impact this has on sub agent work.

Фото профиля Rehan
Rehan1 год назад

Sounds interesting

Фото профиля Yuanhao
Yuanhao1 год назад

My vibe flow: Sonnet 4.5 + ultrathink 🫣

Фото профиля Bilal Bakr
Bilal Bakr1 год назад

So its like the Claude 4.5 lite version?

Фото профиля Sri Agung
Sri Agung1 год назад

Haiku > Sonnet, fr/fr

Фото профиля Mr. Code NJ
Mr. Code NJ1 год назад

It's fast but not as smart. I prefer accuracy over speed.

Фото профиля KLX
KLX1 год назад

I just tried to do a pretty simple financial task Sonnet 4.5 does fine and it basically told me I’m a criminal and it can’t help me 💀😂 it seems to either lack a very basic understanding of the world or is WAY too overcautious, or both?

Похожие видео

This workflow will save you thousands with Claude. Run Opus 5.5, Sonnet 5.5, Haiku 5.5, and Fable 5.1 together, and stop spending Opus tokens on work that doesn't need Opus. The entire idea in one line: Opus plans, Sonnet edits, Haiku reads, Fable reviews at decision points. Roles, broken down: - Opus 5.5, high effort, owns the plan and reviews the final code - Sonnet 5.5, medium effort, is the worker: edits files, runs tests - Haiku 5.5, low effort, splits into explorer (searches and reads the codebase) and researcher (pulls docs). it's the first Haiku with an effort setting - Fable 5.1, set with /advisor fable. Opus calls it at decision points, and it gets the full transcript each time Three moments Opus tends to call it: → before committing to a plan: is this the right approach? → the same error shows up again: is this going nowhere? → before marking the task done: did something get skipped? Why Haiku only reads: Anthropic's launch post says Sonnet 5.5 and Opus 5.5 are still the better choice for complex agentic coding (Terminal-Bench 4.0: 39.2% for Haiku 5.5, 70.6% for Sonnet 5.5) and that Haiku 5.5 fits narrowly scoped subagent work. so it gets the lookups, not the edits. Anyone still running one model for everything is paying $4 per million input tokens to check whether a file exists. Haiku 5.5 does it for $0.10. Drop this into Claude Code 👇 "Rebuild my Claude Code setup around this structure: Confirm Claude Code is v2.1.293 or later, so the haiku alias resolves to Haiku 5.5. If it isn't, stop and tell me to run claude update. Look through ~/.claude/agents and .claude/agents for subagents already covering explorer, worker, and researcher. Only create new ones for roles that are missing. Set model: haiku, effort: low on explorer and researcher, with no Edit or Write tools. Set model: sonnet, effort: medium on worker. If an existing subagent is locked to a different model, leave it as is and just list it. Name the explorer subagent Explore so it overrides the built-in one, which otherwise runs on my main model. In ~/.claude/settings.json, set advisorModel to fable, and set effortLevel to high for claude-opus-5-5 under modelSettings. A top-level effortLevel in user settings doesn't apply to Opus 5.5. Check for anything disabling the advisor: CLAUDE_CODE_DISABLE_ADVISOR_TOOL, DISABLE_TELEMETRY, or anything blocking feature-flag fetches. Also check CLAUDE_CODE_EFFORT_LEVEL, which overrides subagent effort settings, and CLAUDE_CODE_SUBAGENT_MODEL_FORCE, which makes Claude Code ignore subagent model fields. Report what you find. Don't change any of it yet. Add one line to ~/.claude/CLAUDE.md: consult the advisor before a big plan, when the same error shows up twice, and before marking a long task done. Show every change as a diff first. Wait for my go-ahead before touching anything."

Alvaro Cintas

58,450 просмотров • 1 месяц назад