What Anthropic actually builds
Anthropic is the American company behind Claude, and five models are generally reachable today: Fable 5, Opus 5, Sonnet 5 and Haiku 4.5, plus Mythos 5 which is invitation-only. Its real strength sits in code and long agentic work. It has never shipped an image or video model at all.
- four public models
- 1M token context window
- no image or video model
Last checked: This is a reference page. It is re-checked against the vendor sources and updated when a new version ships. What changed
Products
The things you actually sign in to, rather than the model names.
Model families
Four model lines, each at a different point on the intelligence to price curve.
Versions
| Versions | Released | Status | Model families |
|---|---|---|---|
| Claude Opus 5 | 24 July 2026 | current | Claude Opus |
| Claude Fable 5 | 9 June 2026 | current | Claude Fable |
Three axes that keep getting conflated
Most confusion about Anthropic comes from treating three separate things as one. Claude is a product, the place you sign in and talk. Opus is a model family, a line with a fixed position on the intelligence to price curve. Opus 5 is a version, one set of weights with its own release date and price. When somebody asks whether Claude beats GPT, it is genuinely unclear which axis they mean.
This page keeps the three apart, and so does every other page in this section.
Where it is serious, and where it is not
Anthropic is the only large lab with effectively no image, video or music model. We do not file that as a weakness, because it reads as a choice: the focus has stayed on text, code and agents. The practical consequence for you is simple. Send graphics and video work to Google or ByteDance, and keep Anthropic for code and analysis.
A documented detail that gets missed
From Opus 4.7 onward the tokenizer changed, and the same text now produces roughly 30 percent more tokens. The million-token window is nominally identical, but the amount of real text that fits is smaller than on earlier generations. If you have costed a service on token throughput, migrating to a newer generation is not just a change in unit price.
What it is good at
- The strongest option today for coding and multi-step agentic work
- A one million token context window across all four public upper-tier models
- Shipped simultaneously on Amazon Bedrock, Google Cloud and Microsoft Foundry, which for companies means a contract and a stated data region
- Relatively candid documentation: reliable knowledge cutoff and training data cutoff are listed as separate dates
Where it falls short
- No image, video or music generation at all. For that work it is the wrong choice.
- Its most expensive model, Fable 5 at $50 per million output tokens, costs several times the cheaper end of the market.
- Mythos 5 is invitation-only with no self-serve signup, so in practice it is not available to you.
- Blocked from Iran, and Iranian cards do not work either. There is no official, painless route.
Our take
If the job is code, or an agent that has to run unattended for hours, Anthropic is the first place we look today. If the job is images or video, there is nothing here for you and looking is a waste of an afternoon.
Questions people actually ask
What is the difference between Fable 5 and Opus 5
Fable 5 is more expensive and slower, and it is positioned for long-running agents at $10 in and $50 out per million tokens. Opus 5 is half that price ($5 and $25) and is presented for complex agentic coding and professional work. Opus 5 also carries a later knowledge cutoff.
Is Claude Code the same thing as Claude
No. Claude is a chat app; Claude Code is a command line tool that works inside a code repository. The model underneath can be the same, but what they do is not.
Sources
- Anthropic model overview vendor source 12 August 2026
- Anthropic news vendor source 12 August 2026