Claude Fable 5 API 2026: Anthropic's New Flagship Model, Pricing, and CodeFast Claude Max API
Anthropic announced Claude Fable 5 on June 9, 2026: the strongest generally available Mythos-class model with 1M context, adaptive thinking, and $10/$50 MTok pricing. This guide compares Fable 5 software engineering, vision, safety classifiers, and official costs with CodeFast Claude Max API's $0.15 per 1M token package approach.
· CodeFast Team
What is Claude Fable 5?
Anthropic announced Claude Fable 5 on June 9, 2026. It is positioned as a Mythos-class model made safe for general use. Developers can use it on the Claude API with the `claude-fable-5` model ID. It targets 1M context, 128K max output, adaptive thinking (always on), and a clear lead over previous Claude models on long-running autonomous tasks.
The main message is clear: Fable 5 is the most capable model Anthropic has ever made generally available. It targets state-of-the-art performance in software engineering, knowledge work, vision, scientific research, and long-running agent tasks. The longer and more complex the task, the larger Fable 5's lead over previous models.
Fable 5 vs Mythos 5 vs Opus 4.8
Claude Mythos 5 is built on the same underlying model as Fable 5; the difference is the safety layer. Fable 5 is protected by classifiers for general users; Mythos 5 is offered in limited access through Project Glasswing to cyber defenders and infrastructure providers with some safeguards lifted. For most developers, Fable 5 is the right choice.
Model snapshot - June 9, 2026
Claude Fable 5 | claude-fable-5 | $10 / $50 MTok | 1M context | 128K output | adaptive thinking (always on)
Claude Mythos 5 | claude-mythos-5 | $10 / $50 MTok | 1M context | 128K output | Project Glasswing only
Claude Opus 4.8 | claude-opus-4-8 | $5 / $25 MTok | 1M context | 128K output | adaptive thinking
Claude Sonnet 4.6 | claude-sonnet-4-6 | $3 / $15 MTok | 1M context | 64K output | extended thinking
Claude Haiku 4.5 | claude-haiku-4-5 | $1 / $5 MTok | 200K ctx | 64K output | extended thinking
June 9, 2026 model comparison snapshot
- Fable 5: highest capability, general API access, safety classifiers active.
- Mythos 5: same underlying model, invitation-only access, cyber-defense-focused use.
- Opus 4.8: lower token cost, still strong but not ahead of Fable 5 on the longest autonomous tasks.
Software engineering and agentic coding
In early testing, Stripe reported that Fable 5 completed a codebase-wide migration in a 50-million-line Ruby codebase in one day that would otherwise take a team more than two months by hand. On Cognition's FrontierCode evaluation, Fable 5 scored highest on difficult coding tasks that meet production-quality code standards. Platforms like Cursor, GitHub Copilot, and Replit also reported clear improvements in long-running coding and agent tasks during early access.
- Large repository migrations, framework transitions, and multi-file refactors.
- Long-running agent sessions in Claude Code and high-autonomy tasks.
- Token efficiency: Fable 5 can complete the same-quality output in fewer turns than previous Claude models.
Vision, memory, and long-context
Fable 5 is positioned as the new state-of-the-art for vision tasks. Examples include extracting precise numbers from detailed scientific figures, rebuilding web app source code from screenshots, and beating Pokémon FireRed with a minimal vision-only harness. On memory, it stays focused across millions of tokens; in Slay the Spire tests, persistent file-based memory improved performance three times more than Opus 4.8.
Safety classifiers and Opus fallback
Because Mythos-class models carry significant risk, Fable 5 is protected by new classifiers. When queries related to cybersecurity, biology/chemistry, or distillation are detected, the response is automatically handled by Claude Opus 4.8 instead, and users are informed. According to Anthropic, more than 95% of sessions involve no fallback at all. The classifiers are deliberately tuned conservatively, so some harmless requests may still fall back to Opus.
- Cybersecurity: exploit development and attack-planning queries fall back to Opus 4.8.
- Biology/chemistry: most biomedical queries currently route to Opus 4.8.
- Distillation: queries aimed at extracting model capabilities are blocked or fall back to Opus.
Official Fable 5 pricing
According to Anthropic's announcement, Claude Fable 5 is priced at $10 per 1M input tokens and $50 per 1M output tokens, less than half the price of Claude Mythos Preview. For comparison, Opus 4.8 standard pricing remains $5 / $25, Sonnet 4.6 is $3 / $15, and Haiku 4.5 is $1 / $5. Fable 5 pricing matches Opus 4.8 Fast Mode.
Official Anthropic pricing snapshot - June 9, 2026
Claude Fable 5 | $10 / 1M input | $50 / 1M output | 1M context | 128K output
Claude Opus 4.8 | $5 / 1M input | $25 / 1M output | 1M context | 128K output
Claude Sonnet 4.6 | $3 / 1M input | $15 / 1M output | 1M context | 64K output
Claude Haiku 4.5 | $1 / 1M input | $5 / 1M output | 200K context
CodeFast Claude Max API public package snapshot
Supported models: claude-sonnet-5, claude-fable-5, claude-opus-4-8, claude-sonnet-4-6, claude-haiku-4-5 and more
1M context: active
Token selection: 5M-500M
Package calculation: $0.15 / 1M token
Delivery: manual approval and account activation
On the official side, Fable 5 carries twice the input cost of Opus 4.8 standard pricing; the CodeFast package approach lowers entry budget.
When does Fable 5 make sense?
- 50M+ line codebase analysis, migration, or multi-stage modernization projects.
- Vision-based tasks: UI rebuilds, scientific figure analysis, multimodal agent flows.
- Long-running autonomous agent work: coding, research, and analysis tasks lasting hours or days.
- For simple chat or low-cost routing, Sonnet/Haiku may still be more sensible.
Using Fable 5 with CodeFast Claude Max API
CodeFast Claude Max API provides access to all supported models, including Fable 5, through the same token package. Base URL is `https://api.codefast.app/claude-api`, and the Anthropic Messages endpoint is `https://api.codefast.app/claude-api/v1/messages`. For Claude Code, use the same address as ANTHROPIC_BASE_URL and set the model to `claude-fable-5`.
PowerShell
$env:ANTHROPIC_BASE_URL="https://api.codefast.app/claude-api"
$env:ANTHROPIC_API_KEY="YOUR_CODEFAST_API_KEY"
$env:ANTHROPIC_MODEL="claude-fable-5"
claude
macOS / Linux
export ANTHROPIC_BASE_URL="https://api.codefast.app/claude-api"
export ANTHROPIC_API_KEY="YOUR_CODEFAST_API_KEY"
export ANTHROPIC_MODEL="claude-fable-5"
claude
curl example
POST https://api.codefast.app/claude-api/v1/messages
model: claude-fable-5
max_tokens: 4096
CodeFast Claude Max API quick setup
On the CodeFast side, the main value is package-based Claude access with a lower entry budget. The public package data lists active 1M context, selectable 5M-500M token amounts, a $0.15 per 1M token calculation, and account activation after manual delivery.
- Direct Anthropic API: for teams that need official console, usage credits, enterprise settings, and direct invoices.
- CodeFast Claude Max API: for developers who want to test Claude access including Fable 5 with a lower budget and token-package model.
Checklist before production
- Use Fable 5 only for hard and long tasks; route simpler work to Sonnet or Haiku.
- Do not auto-fill the 1M context; design summarization, file selection, and cache strategy first.
- Some queries may fall back to Opus 4.8 because of safety classifiers; this is expected behavior.
- Output token cost is 5x input; plan budget separately for agent work that produces long output.
- Before buying, re-check the current token price, minimum selection, and delivery notes on the CodeFast package screen.
Conclusion: new flagship, new cost plan
Claude Fable 5 API is one of the most important Claude updates of 2026. Bringing Mythos-class capabilities to general use sets a new ceiling for software engineering and long agent tasks. But like every flagship model, Fable 5 requires understanding cost planning, routing strategy, and safety classifiers. CodeFast Claude Max API stands out as a strong alternative for developers who want to test and grow Fable 5 usage with a more controlled budget.