Anthropic Models: Every Model, Spec, Price and Release Date

Anthropic is an AI company whose Claude large language models are offered through the Claude API, the Claude apps and major cloud platforms, and its model line-up changed several times in 2026. This hub lists every Claude model in Anthropic's public documentation as of 15 November 2026, with API IDs, release dates, context windows, prices, availability, access restrictions and retirement dates. It is written for developers, buyers, architects and anyone who needs to know which Claude model is which. Models are listed by release date, newest first, an objective order that implies no ranking. Every fact comes from Anthropic's own documentation, cited in the sources list.

Last checked: 2026-11-15.

This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider. It uses Anthropic's public documentation as of November 2026. Model details change, so confirm against Anthropic's current documentation.

Key takeaways

• Five model lines are current. Anthropic's docs list Claude Fable, Mythos, Opus, Sonnet and Haiku models; the four latest general models are Fable 5.2, Opus 5.5, Sonnet 5.5 and Haiku 5.5, and Mythos 5.2 shares Fable 5.2's capabilities under restricted access.
• Context is 1M tokens on every 4.6 and later model. Claude 4.6 and later models list a 1M-token context window and 128K max output; Opus 4.5, Sonnet 4.5 and Haiku 4.5 list 200K and 64K.
• Prices run from $0.10 to $10 per million input tokens. Haiku 5.5 starts at $0.10 input and $0.50 output; Fable 5.1 and Mythos 5.1 are $10 and $50; Batch API requests cost 50% of standard rates.
• Mythos models are restricted. Mythos 5 and 5.1 are for Project Glasswing participants and Mythos 5.2 is for organisations verified through Anthropic verification programmes.
• Retirement dates are published. Active models carry a "not sooner than" retirement date, Anthropic gives at least 60 days' notice for publicly released models, and Sonnet 4.5 retires on the Claude API on 30 November 2026.

What is Anthropic?

Anthropic develops the Claude family of large language models. Its documentation groups the current models into five lines, Fable, Mythos, Opus, Sonnet and Haiku, and offers them through the Claude API, the Claude apps and several partner clouds. All Claude models are proprietary: Anthropic publishes no model weights and states no parameter counts. Claude 4.6 and later models use undated API identifiers such as claude-opus-5-5, while earlier models carry a date snapshot in the identifier.

The main table below lists the Claude models released in 2026. Earlier models that are still active, such as Claude Opus 4.5, Sonnet 4.5 and 4.6, and Haiku 4.5, appear in the pricing and deprecations tables. Anthropic publishes a "not sooner than" retirement date for each active model and gives at least 60 days' notice before retiring a publicly released model. Facts are as of 15 November 2026.

Current Claude models at a glance

As of 2026-11-15. Source: Anthropic models overview, model deprecations page and release notes. Models are listed by release date, newest first. Category is the Anthropic model line.

The models overview states that all current models accept text and image input, produce text output, and support tool use and multilingual use. The models overview lists default effort "high" on Fable 5.2 and Sonnet 5.5 and "medium" on Opus 5.5 and Haiku 5.5. No Claude model is published with open weights, and Anthropic does not state parameter counts. See Open-Weight vs Closed Models Explained.

Key facts for the main Claude models

As of 2026-11-15. Identical fields for every model; "Not stated by Anthropic" means not stated by Anthropic on the pages checked (2026-11-15).

Default effort and relative latency are the descriptors Anthropic uses on the models overview in place of positioning statements for most models. Mythos rows share the Fable 5.2 or Fable 5.1 price and context, but access is restricted, as set out below. Max output of 128K applies to every model in the table; the 4.5 generation, still active, lists 64K.

Features stated across current models

• Input and output: text and image input, text output, on every current model.
• Tool use and multilingual use: supported on all current models, per the models overview.
• Batch API: 50% of standard rates; not combinable with fast mode.
• Fast mode: research preview on Opus 5.5 from 22 September 2026 and launched on Fable 5.2 on 9 November 2026; deprecated on Opus 4.8 with removal on 30 November 2026.
• Prompt caching: 5-minute and 1-hour cache writes are priced per model in the pricing table.

Claude model families and naming

• Fable: the line Anthropic describes for demanding reasoning and long-horizon agentic work (Fable 5, 5.1, 5.2). Fable 5 was released to all customers.
• Mythos: restricted-access models. Mythos 5.2 has the same capabilities as Fable 5.2, per the 9 November 2026 release note. An earlier Claude Mythos Preview (claude-mythos-preview) was deprecated on 9 June 2026.
• Opus: general high-capability models (Opus 4 to 4.8, Opus 5, Opus 5.5).
• Sonnet: mid-tier models (Sonnet 4, 4.5, 4.6, 5, 5.5).
• Haiku: the fastest tier by latency on the models overview (Haiku 4.5, Haiku 5.5).
• API IDs: 4.6 and later models use undated IDs such as claude-opus-5-5; older models carry a date snapshot, for example claude-sonnet-4-5-20250929. See LLM Naming and Versioning Explained.

When was each Claude model released?

As of 2026-11-15. Dates are from the Claude Platform release notes, in date order.

The latest releases are Claude Fable 5.2 and Claude Mythos 5.2, on 9 November 2026. Claude Haiku 5.5 (7 October 2026) is the most recent Haiku model, and Claude Sonnet 5.5 (28 September 2026) the most recent Sonnet model.

How much do Claude models cost?

As of 2026-11-15. USD per million tokens (MTok), standard (global) processing. Batch shows input / output. Cache write shows the 5-minute and 1-hour write prices. Source: Anthropic pricing page.

• Long context: Claude 4.6 and later models, except Haiku 5.5, charge the full 1M-token window at standard rates. Haiku 5.5 is the one model tiered by prompt length, at 100,000 tokens.
• Fast mode: up to 2.5x output speed at premium rates: Fable 5.2 and Mythos 5.2 at $16 / $80, Opus 5.5 at $8 / $40, Opus 5 and Opus 4.8 at $10 / $50. Fast mode is not available with the Batch API.
• Data residency: on Claude 4.6 and later, inference_geo set to "us" applies a 1.1x multiplier to all token categories; "global" is the default at standard pricing.
• Tokeniser: the pricing page notes that the 4.7 and later tokeniser produces about 30% more tokens for the same text, so per-token prices do not compare directly across that boundary.

See Model Pricing Explained and How to set spend limits on LLM API keys.

Where are Claude models available?

• First-party: the Claude API and the Claude apps.
• Cloud platforms: the release notes list Opus 5, Fable 5.1, Opus 5.5, Sonnet 5.5 and Haiku 5.5 on the Claude API, Amazon Bedrock, Claude Platform on AWS, Google Cloud (Google managed) and Microsoft Foundry. Fable 5.2 launched on all of these plus Claude Models on Google Cloud (Anthropic managed), launched 22 October 2026.
• Regional processing: on Bedrock and Google Cloud, regional and multi-region endpoints for Sonnet 4.5, Haiku 4.5, Opus 4.5 and later carry a 10% premium over global endpoints.
• Retirement schedules on partner clouds: Amazon Bedrock and Claude Models on Google Cloud (Google managed) set their own retirement dates, which can differ from Anthropic's.

Fable 5 was released to all customers on 9 June 2026; Mythos models are listed on the same platforms only for organisations that meet the access conditions in the next section. Anthropic's documentation states no open-weight download for any Claude model, so self-hosting is not an option.

Restricted-access Claude models

As of 2026-11-15. Source: Anthropic models overview, release notes and zero data retention page.

Fable 5, 5.1 and 5.2 are generally available but are also Covered Models that require 30-day data retention.

What is the context window of Claude models?

Every Claude model from Opus 4.6 and Sonnet 4.6 onward lists a 1M-token context window and 128K-token max output. Claude Opus 4.5, Sonnet 4.5 and Haiku 4.5 list 200K context and 64K output; the retired Opus 4 and 4.1 listed 200K and 32K. Haiku 5.5 is the one model whose price changes with prompt length, at 100,000 tokens; every other 4.6 and later model bills the full 1M window at standard rates. See Context Windows Explained.

Is any Claude model open source?

No. Anthropic publishes no Claude model weights; all models are proprietary and accessed through Anthropic or partner cloud APIs and apps.

Model cards and system cards

As of 2026-11-15.

The models overview acts as Anthropic's specification page: it gives API IDs, context window, max output, knowledge cutoff, default effort, relative latency and price for each current model. See Model Cards and System Cards Explained.

Benchmarks

As of 2026-11-15. Provider-reported and independent results are kept in separate columns.

Anthropic's models overview describes the latest models by relative latency and default effort rather than benchmark scores. See How to Read LLM Benchmarks.

Deprecations and retirements

As of 2026-11-15. Source: Anthropic model deprecations page. Anthropic gives at least 60 days' notice before retiring a publicly released model.

Fast mode on Opus 4.8 is deprecated with removal on 30 November 2026. Every other active model carries a "not sooner than" retirement date, for example 2027-11-09 for Fable 5.2. See the AI Model Retirement Guide.

Data handling and security

• Training use: Anthropic's zero data retention page states that retained data is never used for model training without the customer's express permission.
• Default retention: conversation content is not retained by default except for Covered Models, which require 30-day retention; flagged content may be retained for up to 2 years.
• Zero data retention (ZDR): enabled per organisation through Anthropic sales. Covered Models (Fable 5, 5.1, 5.2 and Mythos 5, 5.1, 5.2) are not available under ZDR unless expressly authorised. Batch processing, the Files API and the code execution tool are among the features marked not ZDR-eligible.
• Restricted models: Mythos 5.1 and 5.2 require 30-day retention and are not available under ZDR unless expressly authorised; Mythos 5 is a Covered Model with 30-day retention.
• Key handling: generic controls apply to every provider's API: keep keys out of code, scope and rotate them, and set spend limits.

See AI Data Retention and Deletion and the Claude for Enterprise Security Guide.

How to choose a Claude model: factual criteria

• Access: Mythos models need Glasswing participation or verification; all other current models are generally available.
• Retention requirements: if the organisation runs under ZDR, Fable and Mythos models need express authorisation.
• Price per token: compare input, output, cache and batch rates in the pricing table, allowing for the 4.7 and later tokeniser.
• Context: 1M on 4.6 and later, 200K on the 4.5 generation.
• Lifecycle: check the retirement date and replacement before building on an older ID.
• Platform: confirm the model is listed in the chosen cloud's catalogue and region.
• Batch and fast mode: batch requests are 50% of standard rates but cannot use fast mode; fast mode is priced at a premium on the models listed.
• Data residency: inference_geo "us" adds a 1.1x multiplier on Claude 4.6 and later.

Practitioner checklist

• Pin production code to an explicit API ID and record it in the AI inventory.
• Review the model deprecations page on a schedule to catch deprecations.
• Move workloads off claude-sonnet-4-5-20250929 before 30 November 2026.
• Check whether the chosen model is a Covered Model before relying on ZDR.
• Re-estimate cost when moving across the 4.7 tokeniser boundary.
• Confirm partner cloud retirement dates separately from Anthropic's.
• Store Claude API keys in a secrets manager and set spend limits. See the LLMjacking Guide.

Standards and references

• Anthropic: Models overview (accessed 2026-11-15)
• Anthropic: Pricing (accessed 2026-11-15)
• Anthropic: Model deprecations (accessed 2026-11-15)
• Anthropic: Claude Platform release notes (accessed 2026-11-15)
• Anthropic: Zero data retention (accessed 2026-11-15)

Related NHI Mgmt Group resources: Claude Code Security Guide, Claude Agent SDK Security Guide, Reasoning Models Explained, LLM Security Guide.