Social posts on October 13 to 15 claim that Claude Fable 5.5 exists, that select partners already have it, and that it briefly showed up as "generally available" in Microsoft Azure's East US 2 model catalog under the ID microsoft-foundry/claude-fable-5-5. Anthropic has not announced it. Its newsroom lists Claude Haiku 5.5 (October 7), Sonnet 5.5 (September 28) and Opus 5.5 (September 22) as the latest launches, and its models overview lists Claude Fable 5.1 as the top model, with no 5.5 version.
This post separates what is claimed from what we could verify, because a launch rumour about a top lab spreads fast and the details people repeat ("generally available", "next week") are exactly the parts that are not confirmed. It follows our earlier reporting on reports that Anthropic weighed a new model to counter GPT-6 Astra.
TL;DR
| Claim | Status |
|---|---|
| Claude Fable 5.5 exists | Unconfirmed; not in Anthropic's newsroom or model docs |
| Partners have early access | Claimed in social posts; no statement from Anthropic found |
| Listed as generally available in Azure's East US 2 catalog | Claimed from screenshots and traces; not visible to us; Azure's page we found covers claude-fable-5 only |
Model ID claude-fable-5-5 appears in LiteLLM data | Not in LiteLLM's public price file when we checked |
| Public release "as early as next week" | Analyst speculation attributed in reports; Anthropic gave no date |
| Fable 5.1 is Anthropic's current top model | Confirmed by Anthropic's docs |
What Anthropic's own pages say today
Anthropic's models overview describes a four-model line-up. We read it on October 15.
| Model | Role per Anthropic | Price per million tokens (input / output) | API ID |
|---|---|---|---|
| Claude Fable 5.1 | Demanding reasoning and long-horizon agentic work | $10 / $50 | claude-fable-5-1 |
| Claude Opus 5.5 | Long-running agentic coding and knowledge work | $4 / $20 | claude-opus-5-5 |
| Claude Sonnet 5.5 | Best mix of speed and intelligence | $2 / $10 | claude-sonnet-5-5 |
| Claude Haiku 5.5 | High-volume, latency-sensitive tasks | From $0.10 / $0.50 | claude-haiku-5-5 |
All four list a 1M-token context window, 128K max output and a June 2026 knowledge cutoff. The docs advise starting with Opus 5.5 for most workloads and using Fable 5.1 "for demanding reasoning and long-horizon agentic work, or when your evals on Claude Opus 5.5 at higher effort still fall short." Anthropic's Opus 5.5 announcement says it performs at the level of Fable 5.1 on most work and costs 40% less to run than Opus 5.
Notice what is missing: there is no Fable 5.5 row, no migration guide for it and no newsroom post. A model ID that does not exist in the docs cannot be called generally available by the lab that makes it.
Where the rumour comes from
The claims, as circulated by several X accounts, are:
- Partners "got access to the upcoming Claude Fable 5.5."
- The model appeared in Azure's East US 2 model catalog with a
GenerallyAvailablestatus. - The ID
microsoft-foundry/claude-fable-5-5appeared in traces flagged by a bot and in LiteLLM's GitHub data for Azure, with a description that looked copied from Fable 5.1. - Some posts predicted a launch "tomorrow" or "as early as next week."
The copied description is the important detail. If a catalog entry reuses another model's text, it may be a placeholder or a pre-staged record, which is a common pattern before launches but also what a mistaken or test entry looks like. It does not show that the model is available to anyone.
What we checked
LiteLLM's public price file. We downloaded the main-branch model price list on October 15 and searched for Fable entries. It contains claude-fable-5 and claude-fable-5-1 across Anthropic, Bedrock, Azure AI, Vertex AI, Databricks, GitHub Copilot and others, and there is no fable-5-5 entry. For comparison, the Azure AI entry for Fable 5.1 lists $10 per million input tokens and $50 per million output tokens, a 1M-token input limit and 128K output, matching Anthropic's docs. That does not disprove the claim: an entry could have been added and reverted, or exist in a branch or a different file. It does mean that anyone citing LiteLLM as proof should link the commit.
Azure's catalog. The Azure AI Foundry catalog page we could find is for claude-fable-5, which describes Fable 5 as the best generally available model for coding and agents. We could not see a Fable 5.5 listing. Catalog entries vary by region and subscription, and Microsoft can change them without notice, so a screenshot is not reproducible evidence.
Anthropic's channels. We found no newsroom post and no docs entry. We did not find a statement from Anthropic or Microsoft confirming partner access.
Why the details matter even if a model is coming
Anthropic's recent cadence makes a new model plausible: Opus 5.5, Sonnet 5.5 and Haiku 5.5 landed within about two weeks, and the 5.5 generation has so far skipped the top tier. Fable 5.1 is the only model in the line that has not had a 5.5 refresh, so a rumour of Fable 5.5 fits the pattern. Plausible is not confirmed, though, and three details in the rumour deserve caution:
- "Generally available" in a catalog is a status field, not a launch. Cloud catalogs sometimes show entries before a model is orderable.
- "Partners have it" could mean evaluation access under NDA, which Anthropic does not usually announce.
- IPO timing speculation is an inference by commentators. Our reporting on Anthropic's business figures is in the Q3 revenue claimed-versus-verified post; none of that confirms a model date.
Timeline of the 5.5 generation, from Anthropic's own pages
Putting the confirmed launches in order shows why the rumour sounds believable and why it still needs a source.
| Date | Confirmed release (Anthropic newsroom) | What Anthropic said |
|---|---|---|
| September 22, 2026 | Claude Opus 5.5 | Performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5 |
| September 28, 2026 | Claude Sonnet 5.5 | A clear upgrade over Sonnet 5 that runs 30% faster and costs up to 30% less for most work |
| October 7, 2026 | Claude Haiku 5.5 | The fastest, cheapest and most capable small model yet, for high-volume, cost-sensitive work |
| October 13 to 15, 2026 | (none) | Social posts claim a Fable 5.5 exists; no Anthropic post |
Three of the four tiers have a 5.5 version. The remaining tier, Fable, is where Anthropic says to go "when your evals on Claude Opus 5.5 at higher effort still fall short." That gap is why commenters expect a Fable 5.5, and it is also why the rumour is easy to invent: anyone who knows the pattern can predict the next ID. A predicted ID appearing in a catalog is therefore weak evidence unless the entry carries real content, such as a distinct description, pricing or limits that differ from Fable 5.1. In the claims we saw, the description reportedly matched Fable 5.1, which points to a placeholder.
The retirement dates on Anthropic's overview page also carry a small hint about cadence. Fable 5.1 is committed "not sooner than September 1, 2027," while Opus 5.5 is "not sooner than September 22, 2027," Sonnet 5.5 "not sooner than September 28, 2027," and Haiku 5.5 "not sooner than October 7, 2027." Each date is roughly a year after the model's launch, which suggests Fable 5.1 shipped around early September 2026. A refresh a little over a month later would be fast but not out of line with how quickly the other 5.5 models arrived. That is inference from published dates, not a prediction we can stand behind.
How to verify a new model yourself
- Query the Models API. With a valid key, list models and look for a new ID. If
claude-fable-5-5returns a model object, it is available to your organisation. - Read the docs page, not a screenshot. Anthropic's models overview lists every current model, its API ID, pricing and retirement date.
- Check cloud provider docs. Bedrock, Vertex AI and Microsoft Foundry each publish their own model ID tables; Anthropic's overview lists these IDs per platform.
- Compare against LiteLLM with a commit link. If someone cites a LiteLLM entry, ask for the commit hash and date.
- Watch for an official post. A real flagship launch comes with a newsroom post, a system card and a pricing line.
What to do if you build on Claude
Do not hard-code claude-fable-5-5 anywhere. If your harness takes a model string from config, keep it configurable, and keep an eval set ready so a new model can be compared against Fable 5.1 and Opus 5.5 on your own tasks the day it ships. For cost planning, Fable 5.1's list price is $10 per million input tokens and $50 per million output tokens; any 5.5 price is unknown. For how the current models compare, see our comparison of Gemini 4 Argon, Opus 5.5, Grok 4.7 and GPT-6 Astra and the Epoch Capabilities Index result for Opus 5.5. If you route across providers, our note on Bedrock model IDs for GLM 5.3 shows how ID tables differ per platform.
What would change this assessment
We will update this post if any of the following appear: an Anthropic newsroom post or docs entry for Fable 5.5, a Microsoft or AWS documentation page listing the ID, a commit in LiteLLM's repository with a dated Fable 5.5 entry, or a statement from Anthropic or a named partner. Until then the accurate summary is: Fable 5.5 is a rumour with a plausible pattern behind it, and nothing we could check shows that it exists.
Related reading
- Anthropic weighs a new model to counter GPT-6 Astra
- Anthropic Q3 revenue: claimed versus verified
- Gemini 4 Argon vs Opus 5.5 vs Grok 4.7 vs GPT-6 Astra
- Opus 5.5 tops the Epoch Capabilities Index
- Bedrock enterprise model IDs for GLM 5.3
- Anthropic's Cyber Verification Program tiers
Checked on October 15, 2026 against Anthropic's newsroom and docs and LiteLLM's public price file; the situation may change quickly.