# Claude Fable 5.2 is now in the eroq chat — what it is and what it costs > Claude Fable 5.2, Anthropic's most capable model, is now in the eroq chat and API. What it does well, how it runs on eroq, its limits and the price per reply. Published 2026-11-14 · eroq.ai — canonical: https://eroq.ai/blog/claude-fable-5-2-on-eroq [Claude Fable 5.2](/models/claude-fable-5-2) is now in the eroq chat. Open the [text studio](/studio/text?model=claude-fable-5-2), pick it in the model list, and ask: it costs 20 credits per reply, the thinking included, with no subscription to Anthropic or to us. The same model answers the eroq API under the id `claude-fable-5-2`. This post covers what the model is, what it is good at, how it runs on eroq and where its limits are. ## What Claude Fable 5.2 is Claude Fable 5.2 is the newest model in Anthropic's Fable line, and the most capable model Anthropic makes available to everyone. It follows Claude Fable 5.1 at a lower price on Anthropic's own API: $8 per million input tokens and $40 per million output tokens, against $10 and $50 for Fable 5.1. Three things set it apart from a typical chat model: - **It always thinks before it answers.** The reasoning step cannot be switched off, only made shorter or longer. That is why it handles questions with many moving parts better than models that answer straight away. - **It is built for long, demanding work.** Anthropic positions it for hard reasoning and long tasks: plans, analysis, long documents, code. - **It is big.** On Anthropic's API it reads up to a million tokens of context and writes up to 128,000 tokens in one reply. On eroq the limits are smaller; see below. ## What it is good at - **Thinking a problem through.** A launch plan with a budget, a comparison between three options, a decision with trade-offs, a step-by-step explanation of something you do not know yet. - **Long writing that holds together.** Scripts, articles, story outlines and product copy that keep the same voice from the first line to the last. - **Editing.** Tighten a draft, change its tone, cut it to a word count, translate it without losing what matters. - **Prompts for the other eroq tools.** Ask it for a five-shot video script with camera moves, a detailed scene for an image, or a voice-over to read with [eroq Voice](/studio/voice), then paste the result into the tool. ## How to use it on eroq 1. Open the [text studio](/studio/text?model=claude-fable-5-2). The link selects Claude Fable 5.2 for you; otherwise pick it in the model list. 2. Write your question or your brief. With no character selected, Claude Fable 5.2 answers as an assistant; pick one of your characters and it plays the scene instead. 3. Send. The price shows before you send, and the conversation stays in your browser. Turn on Private mode to hide your lines from the credit ledger as well. A reply that fails, or comes back empty because the request was blocked upstream, is refunded automatically. ## What it costs 20 credits per reply, flat. You pay the same whether the model thinks for a second or for twenty, and whether the answer is two lines or two pages. Credits are about a cent each at the entry pack, never expire, and stay yours if you cancel a plan. New accounts get 40 free credits, enough for 2 replies from Claude Fable 5.2 to see how it answers your own questions. After that, packs start at $10 and plans at $15 a month; no plan is needed to use the model. ## Its limits on eroq - **Text only.** Claude Fable 5.2 reads and writes text here; it does not look at pictures or files yet. Paste the text you want it to work on into your message. - **16,000 characters per call, history included.** That is a long conversation or a long document, not a book. Start a new conversation when a thread gets long. - **Replies of up to about 900 words.** Ask for the next part if you need more. - **Anthropic's usage policy applies.** Claude declines what that policy rules out, and Anthropic serves the model with its standard data retention rather than zero retention. For roleplay that stays in character and is uncensored within eroq's [acceptable-use policy](/legal/aup), use [eroq RP+](/models/eroq-rp-plus) at 3 credits per reply. ## From the API Claude Fable 5.2 answers the same OpenAI-shaped endpoint as every eroq chat model, with the same key: ```bash curl https://eroq.ai/v1/chat/completions \ -H "Authorization: Bearer $EROQ_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "model": "claude-fable-5-2", "messages": [ { "role": "user", "content": "Give me three names for a ceramics shop, and say why each one works." } ] }' ``` Add `"stream": true` to receive the reply as server-sent events. Without a `system` turn or a `context`, the model answers as an assistant; `max_tokens` defaults to 1,200 and `temperature` is ignored, since the model sets its own sampling. Every parameter is in the [chat API reference](/docs/chat). ## FAQ ### Can I use Claude Fable 5.2 without a Claude subscription? Yes. On eroq you pay per reply in credits, 20 credits each, with no monthly plan required. New accounts get 40 free credits to try it. ### How much does a Claude Fable 5.2 reply cost on eroq? 20 credits, flat: the thinking before the answer is included, and the price shows before you send. A reply that fails is refunded. ### Is Claude Fable 5.2 uncensored on eroq? No. Anthropic's usage policy applies to every request. For uncensored roleplay within eroq's acceptable-use policy, use eroq RP+. ### Can Claude Fable 5.2 read images on eroq? Not yet: on eroq it answers text. Paste the text you want it to work on into your message. ### What is the model id for the API? `claude-fable-5-2`, on `POST /v1/chat/completions`, with your eroq API key.