AI Models. Claude (Anthropic)
Generalist / documents / analysis / agents

Claude (Anthropic)

Fable 5 temporarily unavailable, Mythos-class, 1M context

Claude Fable 5 (released 06/09/2026) was introduced as the first generally available Mythos-class model — Anthropic's most capable public model, SOTA in coding, knowledge work and vision, with the lead growing with task length. Safeguards: queries in cyber/bio/chemistry/distillation areas are handled by Claude Opus 4.8 (fallback in <5% of sessions). Update 06/13/2026: Anthropic suspended access to Fable 5 and Mythos 5 after a 06/12/2026 US export-control directive covering foreign-national access; to comply, Anthropic disabled the models for all customers and says it is working to restore access. Until access returns, treat Fable 5 as a historically described model, not a production-ready choice. Below it: Opus 4.8 (05/28/2026) and Sonnet 4.6 — effort control, Fast Mode, Extended Thinking.

Verified: 2026-06-13

Purchase decision (when to choose / when to avoid)

Choose if...

  • You need the highest quality on long, autonomous tasks — Fable 5 only after access is restored.
  • You have long documents/contracts and want high-quality analysis (long context).
  • You're building agents and tools (tool use) and care about safety and stability.
  • You need enterprise options (e.g. data residency) and transparent materials (system cards).

Avoid if...

  • You need a production-available model right now — Fable 5/Mythos 5 are suspended since 06/12/2026.
  • You have very high volume and token cost is critical — consider cheaper models/hosting.
  • You need open-weight and full weight control (on-prem without vendor lock-in).
  • You require Zero Data Retention — Fable 5 mandates 30-day retention (choose Opus/Sonnet).

Cost in practice (scenarios)

1-3 people (marketing/ops) — quick start

Typically: Pro/Team plan + occasional API use for automation.

  • copy + emails + document analysis
  • a few longer files per week
Team of 5-20 — daily work

Usually: Team/Enterprise (SSO, control, higher limits).

  • standard company workflows
  • governance and retention requirements
Automations (API) — high volume

Cost depends on context length; with long context it grows non-linearly.

  • RAG / document analysis
  • cache and context compression matter
These are estimates/scenarios (not an invoice). Actual cost depends on context length, number of users, limits and retention policies.

Deployment / data / enterprise

Deployment channels

  • Claude (app) — quick start for teams
  • Claude API — integrations and automations; Fable 5 (claude-fable-5) is currently unavailable
  • Amazon Bedrock / Vertex AI / Microsoft Foundry (depending on region and offering; re-check Fable 5 availability after the 06/12/2026 suspension)

Data policy

Training on data
API/Enterprise: not by default (check terms in sources).
Retention
Default API retention is time-limited; Fable 5/Mythos 5: mandatory 30-day retention (safety classifiers), data not used for training.
Data residency
Data residency options available (e.g. inference_geo) — depending on plan/service.
In practice: for sensitive data choose API/Enterprise + retention policies + DPIA.

Enterprise readiness

Admin
Account and policy management in team/enterprise plans.
SSO/SCIM
Depending on plan (Team/Enterprise).
Audit
Depending on plan (Team/Enterprise).
DPA
Available in business/enterprise variants (check terms).
Certifications
Depending on product and agreement (check sources).
Best choice when compliance and control matter more than lowest cost.

Best use cases

  • the most demanding, long-horizon agentic tasks (coding, research, analysis) — after access returns: Claude Fable 5
  • long document analysis (reports, contracts, specifications) — context up to 1M tokens
  • projects where safety and transparency are a priority (system cards, safety classifiers, retention policies).

Strengths

  • Fable 5: SOTA on most benchmarks (incl. FrontierCode), long autonomous tasks, vision, long-context memory.
  • Long context (1M tokens) and strong safety/capability evals in system cards; mature API and data residency options.
  • Opus 4.8 / Sonnet 4.6 as cheaper tiers: Extended Thinking, effort control, Fast Mode.

Weaknesses / risks

  • Access to Fable 5 and Mythos 5 has been suspended since 06/12/2026 after a US export-control directive; Anthropic says other models are unaffected.
  • Cost if restored (Fable 5: $10/$50 per 1M tok.); no weights (vendor lock-in); classifiers can be overcautious (fallback to Opus 4.8 even on benign requests).
  • Fable 5 required 30-day data retention (safety classifiers) — no ZDR; availability and terms after restoration need a fresh Anthropic docs check.

Current models (examples)

  • Claude Fable 5 (06/09/2026) — Mythos-class, most capable public model; currently unavailable after the 06/12/2026 US government directive.
  • Claude Mythos 5 (06/09/2026) — same model with some safeguards lifted; currently also suspended.
  • Claude Opus 4.8 (05/28/2026) — Opus-class flagship, coding/agents, effort control, Fast Mode, Extended Thinking, 1M context.
  • Claude Sonnet 4.6 (02/17/2026) — upgrade in coding/agents/long-context, 1M token context.

Alternatives (if this model doesn't fit)

OP
OpenAI
wider ecosystem and API tooling
GE
Gemini
GCP/Vertex and multimodality in Google ecosystem
MI
Mistral
EU + self-host options (depending on model)