Claude 5.2: Opus 5.2, Fable 5.2 and What You Can Use Today

What Claude models exist right now?

Anthropic's own models overview lists four current models. Everything else on this page is either legacy or unannounced.

Model	Released	Official price (input / output per 1M)	API model ID	Positioning
Claude Fable 5.1	September 1, 2026	$10 / $50	claude-fable-5-1	Demanding reasoning and long-horizon agentic work
Claude Opus 5	July 24, 2026	$5 / $25	claude-opus-5	Complex agentic coding and enterprise work
Claude Sonnet 5	June 30, 2026	$2 / $10	claude-sonnet-5	Best combination of speed and intelligence
Claude Haiku 4.5	2025	$1 / $5	claude-haiku-4-5-20251001	Fastest model with near-frontier intelligence

Anthropic's own guidance is worth quoting because it settles most "which Claude should I use" threads: start with Claude Opus 5 for most workloads, and move to Claude Fable 5.1 for demanding reasoning and long-horizon agentic work, or when Opus 5 at higher effort still falls short on your own evals.

Two details from the Fable 5.1 release notes matter for cost planning. Fable 5.1 launched at $10 / $50 per MTok, the same as Claude Fable 5, not as a cheaper tier — the cost improvement is in cache reads, which were cut to $0.25 per MTok, plus 1M-token context and 128k max output by default. And Sonnet 5's introductory $2 / $10 pricing became permanent on August 10, 2026, cancelling a scheduled increase to $3 / $15.

Is Claude 5.2 released? What is confirmed and what is not

Item	Status as of September 20, 2026
Official Anthropic announcement of Opus 5.2 or Fable 5.2	None
Confirmed model ID	None published
Confirmed price	None published
Confirmed release date	None published
Confirmed context window, output limit or benchmark sheet	None published
Reuters report that Anthropic is considering a further model release	Reported September 18–19, 2026 — real reporting, no model name
Community routing reports naming Opus 5.2	Circulating since roughly September 14, 2026 — not vendor confirmation
Latest model you can call today	Claude Fable 5.1

Treat the difference between those rows as the whole point of this page. A model name appearing in a product slug, a gateway response or an aggregator's model list does not establish that Anthropic has released it, and it does not establish that the endpoint behind that name is serving that model.

Why the market expects a 5.2 generation

The expectation is not built on an Anthropic promise. It is built on three things that happened within a week.

Reuters reported a real internal debate. On September 18–19, Reuters reported that Anthropic is evaluating whether to release an additional model, framed around OpenAI's GPT-6 Astra gaining ground in enterprise accounts and around the company's upcoming IPO story. The report says safety evaluations are running in parallel, and it does not name the model. That is the strongest single signal available, and it is still a report about internal discussions rather than a launch plan.

The timing collides with Anthropic's own public position. On September 12, Dario Amodei published a long essay urging the industry to slow the pace of capability releases. A company publicly arguing for restraint while reportedly preparing another frontier release is exactly the kind of tension that generates speculation — and it is also why some of that speculation is likely to be wrong in either direction.

Spending share gives the commercial motive. Enterprise spend trackers cited alongside the Reuters report put OpenAI's Astra at roughly 13% of tracked enterprise AI spend against roughly 8% for Claude Fable. A gap of that shape is the standard argument for shipping a counter-release quickly rather than waiting for a scheduled generation.

What the community is actually seeing

Everything in this section is unverified. It is useful as a map of what people are watching, not as evidence of a release.

• Routing reports naming Opus 5.2. Since roughly September 14, developers have reported Claude Code and cloud-platform requests being served under an Opus 5.2 label. A routed response that says Opus 5.2 tells you the gateway or client was configured to report that string — the model field is filled in by the serving party, and a relay can set it independently of what generates the tokens. It is a genuine signal that something changed upstream in those pipelines, and it is not proof that Anthropic shipped a model under that name.
• A claude-opus-5-2 slug. Screenshots of claude-opus-5-2 appearing in Microsoft Foundry surfaced in the same window. A slug in a catalogue or a console is a product-configuration artifact: it shows someone prepared an entry, not that the entry serves traffic. Both this and the routing reports are consistent with a staged rollout, a partner preview, or a naming rehearsal.
• A skipped Opus 5.1. Several threads argue Anthropic jumped from Opus 5 straight to 5.2 rather than shipping an Opus 5.1. There is no published roadmap either way, and this is the kind of claim that is impossible to falsify until the next release ships with a number on it.
• Fable 5.2 paired with Mythos 5.2. On X, the two names have been appearing together, mirroring the Fable/Mythos 5.1 pattern where Fable is the broadly available version and Mythos is the restricted, higher-risk sibling. The pairing is the community's extrapolation from the existing release shape.
• The internal name Model 2. August risk-report material described an unreleased internal model stronger than Mythos 5, referred to as Model 2 and not intended for external release at the time. If it later reaches customers, the external name would almost certainly be an Opus/Fable/Mythos number rather than Model 2 — which is why an internal codename is the weakest possible input for a public product name.

Our existing verification guide explains the same distinction from the buyer's side: never treat a model's claim about its own identity as proof of which model served a request.

Claude 5.2 pricing: not published, and the Claude API prices you can pay today

No price exists for Opus 5.2, Fable 5.2 or Mythos 5.2. Anthropic has not published one, and anyone quoting a specific number for these models is quoting a prediction, not a price. The three patterns that are plausible on the current record — same price as the tier it replaces, a cache-economics improvement like the one Fable 5.1 shipped, or a lower introductory price — are expectations, not facts.

What you can price today is the current generation. Every figure below comes from APIMaster's live marketplace data for the model ID, filtered to channels that are actually selling, and was checked on September 20, 2026:

Model	Official (input / output per 1M)	Live routes	APIMaster from	Saving	Card
Claude Fable 5.1	$10 / $50	6	$2.8944/M input	up to ~71%	Fable 5.1
Claude Fable 5	$10 / $50	4	$6.5424/M input	up to ~35%	Fable 5
Claude Opus 5	$5 / $25	11	$1.4847/M input	up to ~70%	Opus 5
Claude Sonnet 5	$2 / $10	12	$0.1875/M input	up to ~91%	Sonnet 5
Claude Sonnet 4.6	$3 / $15	11	$0.2813/M input	up to ~91%	Sonnet 4.6
Claude Opus 4.7	$5 / $25	13	$0.45/M input	up to ~91%	Opus 4.7
Claude Opus 4.8	$5 / $25	13	$0.45/M input	up to ~91%	Opus 4.8
Claude Haiku 4.5	$1 / $5	7	$0.09/M input	up to ~91%	Haiku 4.5
Claude Opus 5.2 / Fable 5.2	Not published	0	Reserved — this row gets a number when Anthropic ships one	—	—

Two caveats belong next to that table. First, "from" is the cheapest live route, not the only price: the same model's other channels run all the way up to Anthropic's list price, and one Sonnet 5 channel is currently priced above it. Second, route prices and channel supply move, so the card is authoritative rather than this table.

To use any row: create an APIMaster account, add a small balance (pay-as-you-go starts from $1), generate a key in the console, and set the model string. The reserved row is deliberate: when 5.2 exists as a real model with a real ID and live routes, this page carries its price and you can compare it card-for-card against everything above.

What to use while you wait

The decision is simpler than the rumor cycle suggests.

• You want the strongest Claude available today. Claude Fable 5.1. It is Anthropic's current answer for demanding reasoning and long-horizon agentic work, and it is the tier to benchmark against if Astra is the model you are actually worried about.
• You want the best value on general agentic coding and enterprise workloads. Claude Opus 5 at $1.4847/M input. It is Anthropic's own default recommendation for most workloads, it has the largest and most stable supply of any current Claude tier on APIMaster, and at roughly 70% off official pricing it costs a fraction of a frontier route.
• You are running high-volume, latency-sensitive work. Claude Sonnet 5 at $0.1875/M input, or Haiku 4.5 at $0.09/M for the cheapest tier. Sonnet 5's price is now permanent, so you can plan against it rather than around a promo window.
• You need a specific older behaviour. Opus 4.7 and Opus 4.8 remain available at $0.45/M input on verified routes, which is useful when an evaluation was built against their exact output distribution.

If your real question is "should I wait for 5.2 before building", the honest answer is no. A model that has not been announced has no date, and every week of waiting is a week of not running your own evals — which are the only numbers that will tell you whether 5.2 is worth switching for.

What happens when 5.2 does ship

Three things are worth knowing in advance, because they decide whether a release costs you a migration or a config change.

1. This URL keeps working. This page is the durable tracker for the Claude 5.2 generation. When Anthropic publishes a real model ID, price and limits, they land here and the "not published" rows are replaced — no new URL to find, no redirect to chase.
2. Your key does not change. APIMaster exposes one OpenAI-compatible endpoint, so adopting a new Claude tier is a model string change in your client rather than a new account, a new SDK or a new billing relationship. The same key that calls claude-opus-5 today will call the 5.2 model when a verified route exists.
3. Verify before you migrate. When a new tier appears, run your own evaluation — a failing test, a multi-file refactor, a long-document question — against both the old and new model, and check the route with a fingerprint test before moving production traffic. Benchmarks and routing labels are inputs to that decision, not substitutes for it.

FAQ

Has Anthropic released Claude Opus 5.2 or Fable 5.2?

No. As of September 20, 2026 there is no Anthropic announcement, model ID, price or release date for either name. The newest callable Claude model is Claude Fable 5.1, released September 1, 2026.

Where did the Opus 5.2 and Fable 5.2 names come from?

Mostly from routing reports and product slugs rather than from Anthropic. Developers reported Opus 5.2 labels in routed requests from around September 14, screenshots of a claude-opus-5-2 slug appeared in the same window, and the Fable 5.2 / Mythos 5.2 pairing is the community extrapolating from how Fable and Mythos 5.1 shipped. Reuters has since reported that Anthropic is weighing an additional release, but it did not name a model.

Is there a price for Claude Opus 5.2?

No. Nothing has been published, so any figure you see is a prediction. For reference, the tier it would most plausibly replace, Claude Opus 5, is officially $5 input / $25 output per million tokens and routes on APIMaster at $1.4847 per million input tokens.

Should I wait for Claude 5.2 before building?

No. An unannounced model has no date. Claude Opus 5 and Claude Fable 5.1 are available now with stable IDs and published prices, and the migration to a future tier is a model-string change on the same key rather than a rebuild.

Which Claude model is the strongest today?

Claude Fable 5.1, per Anthropic's own model overview, which positions it for demanding reasoning and long-horizon agentic work and recommends Claude Opus 5 as the starting point for most workloads.

How much cheaper is Claude on APIMaster?

Checked September 20, 2026: Claude Opus 5 is about 70% below Anthropic's official $5/M input, and Sonnet 5, Sonnet 4.6, Opus 4.7, Opus 4.8 and Haiku 4.5 are each about 91% below official. Fable 5.1 routes from $4.46/M input on general channels and $2.89/M on Claude Code channels against an official $10/M. Route prices change, so confirm on the live card.

Will a 5.2 model be more expensive?

Unknown, and anyone who tells you otherwise is guessing. The relevant precedent is mixed: Fable 5.1 launched at the same list price as Fable 5 and delivered its cost improvement through cache reads, while Sonnet 5's introductory pricing was made permanent instead of increasing on schedule.

Sources and verification date

• Anthropic — Models overview: current lineup, API model IDs, official pricing and the Opus 5 / Fable 5.1 positioning guidance. Checked September 20, 2026.
• Anthropic — Claude API release notes: the September 1, 2026 Fable 5.1 and Mythos 5.1 launch at $10 / $50 per MTok with cache reads at $0.25, and the August 10, 2026 Sonnet 5 pricing decision.
• Anthropic — Introducing Claude Opus 5 (July 24, 2026) and Claude Sonnet 5 (June 30, 2026).
• Anthropic — Claude Fable 5 and Mythos 5 and Claude Fable 5.1 and Mythos 5.1.
• Reuters reporting on Anthropic's internal discussions about an additional model release, September 18–19, 2026. The report does not name a model, and Anthropic has not confirmed it; it is cited here as reporting, not as a release.
• Dario Amodei's September 12, 2026 essay on the pace of capability releases, referenced as context for the current speculation.
• Community routing reports and product-slug screenshots circulating from September 14, 2026, and the internal Model 2 reference from August risk-report material. These are treated as unverified throughout.
• APIMaster live marketplace routes and the public Claude pricing table, checked September 20, 2026. Route availability and prices change after publication.

Start with Claude Opus 5 on APIMaster.ai

Opus 5.2 is a rumor. Claude Opus 5 is a model you can call in the next five minutes, at $1.4847 per million input tokens — roughly 70% below Anthropic's official price — with Fable 5.1 available on the same key when a task needs the frontier tier.

Create an APIMaster account, generate a key in the console, point your client at https://apimaster.ai/v1 and set the model to claude-opus-5. Every Claude channel is fingerprint-checked before it is listed, and the same endpoint covers Claude, GPT, Gemini, DeepSeek and Kimi — so when Anthropic does ship 5.2, the upgrade is one string in your config.