Claude Haiku 4.5
Schnelles Claude-Modell für den Alltag
Best for: Everyday questions, short texts and quick answers · fast
- No login required
- Understands images
approx. 912 tokens per response
Anthropic · 4 models
Claude
The Claude models from Anthropic are suited to writing, long documents and programming tasks. Start with Claude Haiku 4.5 right in your browser, no login and no subscription.
Genuine Anthropic models through the official API, no account with the provider. As a guest you get 15,000 tokens with Claude Haiku 4.5. A free account adds 15,000 tokens a day, all models, uploads and saved chats.
What is Claude?
Claude is the AI assistant from Anthropic, a US company founded in 2021 by former OpenAI researchers; Claude has been publicly available since March 2023. Anthropic trains Claude with an approach called Constitutional AI, based on a fixed set of principles, and designs the models for long inputs, careful wording and programming tasks. On ChatX you use Claude Haiku 4.5, Claude Sonnet 5, Claude Opus 5.5 and Claude Fable 5.1 through Anthropic's official API, with no account at Anthropic.
The Claude models on ChatX
Choose the model that fits your task. Fast models are enough for everyday questions, while the larger models pay off for long texts, analyses, and code.
Claude Sonnet 5
Präzise Texte und zuverlässiger Code
Best for: Precise writing, translations and clean code · balanced
- With account
- Understands images
- Web search
approx. 1,823 tokens per response
Claude Opus 5.5
Tiefes Denken für komplexe Arbeit
Best for: Deep analysis and long documents with thinking mode · thorough
- With account
- Understands images
- Web search
- Thinking mode
approx. 3,645 tokens per response
Claude Fable 5.1
Anthropics stärkstes Modell
Best for: Most powerful model for the hardest tasks · thorough
- With account
- Understands images
- Thinking mode
approx. 9,112 tokens per response
New at Claude
Blog →
Claude Opus 5.5 ist günstiger und denkt immer
Anthropic hat am 22. September Claude Opus 5.5 veröffentlicht und damit Opus 5 ersetzt. Zwei Dinge fallen sofort auf: das Modell kostet ein…
Claude Opus 5 und Fable 5.1 im Vergleich
Anthropic hat innerhalb weniger Wochen zwei Modelle erneuert: Claude Opus 5 ersetzt Opus 4.8 zum gleichen Preis, Claude Fable 5.1 bleibt da…
Claude Opus 4.8 bringt mehr Kontrolle in Claude Code
Mit Claude Opus 4.8 legt Anthropic den Schwerpunkt weniger auf einzelne Spitzenwerte und stärker auf verlässlichere Arbeit im Alltag. Das M…
What you can do with Claude on ChatX
Write and edit texts
Claude writes emails, job applications and posts with a clear structure and in your tone.
Work through long documents
With a free account you upload a PDF or Word file and get summaries or answers about specific passages.
Code
Claude explains code, finds bugs and writes functions in common programming languages.
Translate and polish
Claude translates between English, Spanish, German and many other languages and adjusts the tone to the occasion.
Learn and understand
Claude explains difficult topics step by step, with examples and questions that keep you thinking.
Weigh options and decide
In thinking mode, Claude Opus 5.5 and Claude Fable 5.1 reason before answering and sort out arguments and numbers.
Claude compared with the other models on ChatX
Pick Claude when you need well-structured long texts, precise wording or reliable code; Claude has been considered especially strong here for years and handles very long inputs. ChatGPT is the broader all-rounder with more extra features, Gemini and DeepSeek use fewer tokens, and Grok is quick on logic tasks, while Claude doesn't create images, offers web search only on some models and uses more tokens per answer with Claude Opus 5.5 and Claude Fable 5.1. On ChatX you can switch between Claude and the other models in the same chat at any time and simply ask the same question again.
| Model | Strength | No sign-up | Features |
|---|---|---|---|
| ChatGPT | All-rounder for writing, analysis and code | GPT-6 Luna | Web search · Understands images · Thinking mode |
| Claude | Long texts, careful language, code | Claude Haiku 4.5 | Web search · Understands images · Thinking mode |
| Gemini | Fast, large context, understands images | Gemini 3.5 Flash Lite | Web search · Understands images · Thinking mode |
| Grok | Direct tone, quick answers | Grok 4.3 | Understands images · Thinking mode |
| DeepSeek | Math, logic and code at a low price | DeepSeek V4.1 Flash | Understands images · Thinking mode |
All models run in the same chat. Switch any time without losing your history.
What makes Claude different
Anthropic built Claude around safety and reliability from the start. The model is trained with a fixed set of rules that Anthropic calls “Constitutional AI”: instead of having people rate every single answer, Claude learns to measure its answers against these principles. In practice you notice this in a calm, careful style and in the way Claude asks when a task is unclear instead of guessing.
Its second strength is long texts. Claude works through entire contracts, studies or book chapters in one go, keeps track of the context and writes remarkably natural English. Then there is code: since the Claude 4 generation, Claude has been one of the most widely used models for programming work, and on ChatX you choose Sonnet 5 or Opus 5.5 for it.
There are two things Claude doesn't do: it doesn't create images, so for that ChatX has the AI image generator with other models. And it only searches the web on models with the globe icon. For quick everyday questions, Haiku 4.5 is enough, and it runs without signing up.
Costs and free allowance
- Without signing up, you get 15,000 tokens to try it out. That's enough for around 16 answers with Claude Haiku 4.5.
- With a free account, you get 15,000 tokens as a gift every day.
- If you need more, you can buy tokens with no subscription or choose a monthly plan. Billing is based on the model's actual usage.
Privacy: what happens to my input?
Your message goes from ChatX straight to the official API of US company Anthropic, and the answer comes back the same way. You don't need an Anthropic account for this; you use Claude anonymously or with your ChatX account. According to Anthropic's terms (as of November 2026), Anthropic may not train models on content from its API; inputs and outputs are normally deleted within 30 days and stored in the US.
ChatX stores your chats only in your own account. You can delete them one by one or all at once at any time. How Anthropic processes the requests it receives is governed by its terms; you'll find the details in the privacy policy.
Anthropic puts particular weight on safety rules and publishes its principles for Claude openly. Still, as with any AI service: don't enter passwords, health data or other sensitive information you wouldn't want sent to a service in the US.
Claude by Anthropic, explained simply
Claude is developed by Anthropic and is suited to well-structured writing, precise wording and programming tasks. The Claude models are built for long inputs. On ChatX you choose between several Claude models: Claude Haiku 4.5 for everyday use runs without signing up; with a free account you also get Claude Sonnet 5, Claude Opus 5.5 and Claude Fable 5.1, Opus and Fable with thinking mode for deep analyses and long texts, and web search on the models that offer it. Everything runs through Anthropic's official API, with an English interface and billing per reply instead of a monthly subscription. You do not need an account with the provider.
It's this easy
-
1
Enter your question
Type your question or task into the input field. You don't need to log in for this.
-
2
Choose a model
Start with the default model or switch to a more powerful one anytime.
-
3
Keep going
With a free account, you save your chats, upload files, and get free tokens every day.
Frequently asked questions
Which Claude model fits which task?
For everyday use, Claude Haiku 4.5 is enough, and it runs without signing up. With a free account you choose Claude Sonnet 5 for precise writing and code, Claude Opus 5.5 for deep analysis and Claude Fable 5.1 as the most powerful model; Opus and Fable offer thinking mode.
What do the names Haiku, Sonnet, Opus and Fable mean?
Since Claude 3 in 2024, Anthropic has named its model tiers after forms of poetry: Haiku is small and fast, Sonnet is the mid-range and Opus the large line. Fable joined in 2026 as the top model.
Can Claude create images?
No. Claude doesn't generate images itself, but with an account it understands uploaded photos, screenshots and PDF files. You create images on ChatX with the AI image generator, which uses other models, even without signing up.
Can Claude search the internet?
For the Claude models with web search (globe icon in the input field), you turn on search in the chat and get current sources in the answer. Without web search, Claude answers from its training knowledge, which has a cutoff date.
What does thinking mode do for Claude?
In thinking mode, Claude Opus 5.5 and Claude Fable 5.1 work through the problem step by step before answering. That helps with analysis, math and code, takes longer and uses more tokens. Thinking mode is available with a free account.
Does Claude speak English and other languages?
Yes. Claude writes excellent English, handles many other languages at a high level and answers in the language you write in. The ChatX interface is available in English too.
How long can documents be?
Uploading documents requires a free account. Without credit there is a character limit per message (currently 10,000 characters with an account); with credit or a subscription there is no plan limit, and technically up to 200,000 characters per message with an attachment are possible. Large models cost more tokens per answer than Haiku.
How is this different from claude.ai?
claude.ai is Anthropic's own app with an account and a monthly subscription. On ChatX you use the same Claude models through the official API, start without signing up, pay only for actual usage per answer and have models from other providers in the same chat.
The history of Claude
From a research model kept under wraps in 2022, through the public launch in 2023, to the model family with Haiku, Sonnet, Opus and Fable: how Anthropic's safety project became one of the most important AI assistants.
Newest events first
-
Claude 5 and Fable · 2026
-
Produkt
Personal Finance: Claude reads account data
As a beta for paying users in the US, Claude connects to bank, card and investment accounts via Plaid and analyzes spending and contracts. Access is deliberately read-only: Claude can neither transfer money nor invest.
US betaRead-only accessPlaid integration
-
Firma
Anthropic files for an IPO
After a $65 billion funding round in May, Anthropic files with the US Securities and Exchange Commission for its planned IPO. Claude goes from research product to the core of a company ready to go public.
S-1 filed with the SECSeries H in MayIPO planned
-
Modell
The 5.5 generation: Opus, Sonnet and Haiku
According to Anthropic, Claude Opus 5.5 mostly matches Fable 5.1 while answering over 30 percent faster and typically costing 40 percent less than Opus 5. Sonnet 5.5 follows shortly after, and in October Haiku 5.5 arrives as the fastest and cheapest small model. You use Opus 5.5 on ChatX with a free account.
Opus 5.5 on ChatXOver 30% fasterExternally tested by METR
-
Modell
Claude Fable 5.1 and Mythos 5.1
The top model gets a performance and efficiency update. Fable 5.1 is generally available, while the technically identical Mythos 5.1 is offered only through verified programs for cybersecurity and life sciences. On ChatX, Claude Fable 5.1 is the most powerful Claude model.
Fable 5.1 on ChatXMythos for verified users onlyFewer false safety flags
-
Modell
Claude Sonnet 5 becomes the default model
The new Sonnet generation is built more for independent tasks and becomes Anthropic's default model for Free and Pro. In July, Claude Opus 5 follows with performance close to Fable 5 at much lower cost. Claude Sonnet 5 is available to you on ChatX with a free account.
Sonnet 5 on ChatXDefault at AnthropicOpus 5 in July
-
Modell
Claude Fable 5 and Mythos 5: launch, pause, relaunch
For the first time, Anthropic presents two variants of the same model: Fable for everyone and Mythos, with fewer safeguards, for verified security researchers. Three days after launch, Anthropic takes both models offline over their safeguards and brings them back in revised form in early July.
New top linePaused after 3 daysRelaunched Jul 1, 2026
-
Modell
Claude Opus 4.8 with Fast Mode
After Opus 4.7 in April, Opus 4.8 delivers better results in coding, agent tasks and office work, plus a Fast Mode that answers about 2.5 times faster. Anthropic also reports that the model is better at spotting its own mistakes.
Fast Mode ≈ 2.5× fasterBetter self-correction
-
Firma
Dispute with the US Department of Defense
Anthropic refuses to allow Claude to be used for mass surveillance of US citizens and fully autonomous weapons. The Pentagon then classifies the company as a supply chain risk, and Anthropic sues. A court temporarily blocks the measure in March, an appeals court upholds it in September, and as of early November 2026 the dispute is still unresolved.
Usage rules defendedSeveral casesNot yet decided
-
Modell
Claude Opus 4.6: a one-million-token context
Opus 4.6 opens a one-million-token context window in beta, adjusts its thinking effort adaptively and brings Agent Teams in Claude Code as well as features for Excel and PowerPoint. Sonnet 4.6 follows two weeks later with the same large context.
1M tokens (beta)Adaptive thinkingAgent Teams
-
Claude 4 and Claude Code · 2025
-
Firma
Claude comes to Microsoft
Microsoft, NVIDIA and Anthropic form an alliance: Claude is built into Azure, Microsoft 365 Copilot, GitHub Copilot and Copilot Studio, and Microsoft and NVIDIA invest in Anthropic. That makes Claude available on all three major clouds. A few days later, Claude Opus 4.5 is released.
Azure and M365 CopilotAWS, Google, MicrosoftOpus 4.5 on Nov 24
-
Modell
Claude Sonnet 4.5 and Claude Haiku 4.5
Sonnet 4.5 improves coding, computer use and long-running tasks. Two weeks later, Haiku 4.5 arrives as the new small generation for fast, low-cost use. On ChatX, Claude Haiku 4.5 is the guest model you start with, no sign-up needed.
Haiku 4.5 without sign-upFast and low-costSonnet 4.5 on Sep 29
-
Modell
Claude Opus 4 and Sonnet 4: the agent generation
Claude 4 combines thinking and tool use: the model can call tools while it reasons, use several in parallel and remember intermediate results. Claude Code becomes generally available. Because of the increased capabilities, Anthropic activates safety level ASL-3 for the first time.
Claude Code generally availableTools while thinkingSafety level ASL-3
-
Produkt
Web search, Research and integrations
For the first time, Claude can draw on current information from the web, first for paying US users and worldwide from May. With Research, Claude investigates over several steps, and through integrations it connects to external data sources and services.
Web search from MarchResearch in AprilIntegrations in May
-
Modell
Claude 3.7 Sonnet thinks longer when asked
As the first Claude model with hybrid reasoning, 3.7 Sonnet either answers right away or takes more computing time with extended thinking, and the budget for it can be controlled. At the same time, Claude Code launches as a preview, a coding assistant for the terminal.
First thinking modeAdjustable thinking budgetClaude Code preview
-
The Claude 3 family · 2024
-
Forschung
Model Context Protocol: an open standard for AI tools
Anthropic publishes MCP, an open protocol that lets AI assistants access data, tools and business systems. The standard is adopted across the industry and later handed over to a foundation under the Linux Foundation umbrella, which OpenAI, Google and Microsoft are also part of.
Open standardThousands of serversLinux Foundation
-
Modell
Claude 3.5 Sonnet and Artifacts
The mid-range model beats the previous top model, Claude 3 Opus, in many tests while being faster. With Artifacts, Claude gets its own workspace for documents, code and drafts. The October successor introduces computer use, the first time Claude can control the screen, mouse and keyboard.
Better than Opus 3ArtifactsComputer use in October
-
Produkt
Claude officially comes to Europe
The web app, the iOS app and the Team plan become available in Europe, so Claude can now be used there officially. The Android app follows in July and the Enterprise plan for businesses in September.
EU launchiOS and AndroidEnterprise in September
-
Modell
Claude 3: Opus, Sonnet and Haiku
A single model becomes a family in three sizes: Haiku fast and low-cost, Sonnet as the mid-range and Opus as the most powerful model. For the first time, Claude can understand images, and it handles a 200,000-token context by default.
Three sizesUnderstands images200,000 tokens
-
Launch and Claude 2 · 2023
-
Modell
Claude 2.1 with a 200,000-token context
The context window doubles to 200,000 tokens, roughly 500 pages of text. According to Anthropic, Claude 2.1 makes about half as many false statements as Claude 2.0 and supports system prompts and tool use in beta for the first time.
200,000 tokensFewer false statementsTool use beta
-
Modell
Claude 2 and the launch of claude.ai
Claude 2 delivers much better results in coding, math and reasoning and handles 100,000 tokens, after the context window had already grown from about 9,000 to 100,000 tokens in May. claude.ai launches as the public web chat in the US and UK, and in September Claude Pro follows as the first subscription.
100,000 tokensclaude.ai betaClaude Pro in September
-
Produkt
Claude goes public
After a closed test phase with partners such as Notion, Quora and DuckDuckGo, Anthropic offers Claude and the cheaper Claude Instant through chat interfaces and the developer API. Shortly after, Claude also becomes available in Slack.
Claude and Claude InstantAPI for developersSlack integration
-
Founding and research · 2021 to 2022
-
Forschung
Constitutional AI: principles instead of human ratings alone
Anthropic publishes the method that still shapes Claude today: the model evaluates and improves its answers against a fixed set of rules (a kind of constitution) and learns from AI feedback rather than only from human ratings. Anthropic makes the principles themselves public in May 2023.
A constitution for ClaudeLearning from AI feedbackPrinciples published 2023
-
Modell
The first Claude stays under wraps
According to Reuters, Anthropic had already trained a first Claude in early 2022 but did not release it, using it for safety research instead. In late 2022, partners such as Notion and Quora's Poe test the model; knowing about it reportedly pushed OpenAI to release ChatGPT sooner.
Trained internallyPartner tests onlyBefore ChatGPT
-
Firma
Anthropic is founded
Dario and Daniela Amodei found Anthropic together with other former OpenAI researchers. The goal is reliable, steerable and interpretable AI systems, with safety built into the concept from the start. A first major funding round follows in May.
Ex-OpenAI teamFocus on safetySeries A in May
Compiled from public announcements by Anthropic and press reports; updated when new models arrive.
More AI models on ChatX
Last reviewed: 9. November 2026 · Source: Anthropic