Claude Fable 5.2, now on RodiumAi
Claude Fable 5.2 joins the RodiumAi catalog as anthropic/claude-fable-5-2.
It is Anthropic's model for the hardest work: demanding reasoning and long-horizon agentic tasks. Think agent sessions that run for hours, multistep deep research, and analysis carried all the way through to a finished document, spreadsheet, or deck.
Anthropic's own guidance is simple: reach for Fable 5.2 when Claude Opus 5.5 at higher effort still falls short on your evals.
At a glance
Model idanthropic/claude-fable-5-2Context window1M tokensMax output128K tokensInputText, image, documentsOutputTextReasoningAdaptive thinking, always onCapabilitiesStreaming, tool calling, vision, JSON mode, reasoningTiersSpeed: Deep · Quality: Max
What it is good at
Long-running agents. Coding agents and tool-using workflows that need to stay on track for hours, not minutes.
Deep research. Multistep investigations across large sets of sources, with a 1M token window to hold them.
Finished deliverables. Analysis that ends in a real document, spreadsheet, or presentation.
Long documents and codebases. Read whole projects, contracts, or reports in one go.
Image understanding. Charts, screenshots, scanned pages.
Writing and editing in English and French.
It also works out of the box in coding tools that speak the OpenAI or Anthropic API: RodiumAi Code, Claude Code, Cline, and Continue.
Pricing
Per 1M tokens, billed in RODI. USD values are a wholesale reference; live RODI prices are on the model page.
Per 1M tokensInput~ $8.00Output~ $40.00Cached read~ $0.20Cache write (5 min)~ $10.00Cache write (1 h)~ $16.00
Fable 5.2 is a premium model. Use it where depth and autonomy justify the cost, and lean on prompt caching for long agent loops: cached reads cost a fraction of fresh input.
Use it with one API key
RodiumAi serves Fable 5.2 through Anthropic, with AWS and Runware as fallback upstreams at the same rates. You call it with your RodiumAi key, in the format you already use.
OpenAI format (/v1/chat/completions):
from openai import OpenAI
client = OpenAI(
base_url="https://api.rodiumai.io/v1",
api_key="YOUR_RODIUMAI_KEY",
)
response = client.chat.completions.create(
model="anthropic/claude-fable-5-2",
max_tokens=8192,
messages=[
{"role": "user", "content": "Plan our Q1 launch: goals, timeline, risks, owners."}
],
)
print(response.choices[0].message.content)
Anthropic format (/v1/messages):
curl https://api.rodiumai.io/v1/messages \
-H "x-api-key: YOUR_RODIUMAI_KEY" \
-H "content-type: application/json" \
-d '{
"model": "anthropic/claude-fable-5-2",
"max_tokens": 8192,
"messages": [{"role": "user", "content": "Summarize this report in 10 bullet points."}]
}'
Good to know:
max_tokensis required.Streaming works on both endpoints.
Fable 5.2 does not accept
temperature,top_portop_k. RodiumAi removes them from your request, so existing code keeps working.
Get started
Sign up or log in on rodiumai.io.
Top up RODI with Mobile Money in Billing.
Create a key in API Keys.
Call
anthropic/claude-fable-5-2.
Full specs and live prices: Claude Fable 5.2 on RodiumAi.


