1:1 with a GTM engineer
What Got Leaked, What It Reveals, and What It Means for GTM
Source: gist.github.com/gsans/b3007997f8900003c8ff58125a45e15e
Leak published: June 11, 2026 by gsans
Relevance: GTM teams using Claude Code for outbound, research, and content
Someone leaked the Claude Fable 5 system prompt in full. It names the new model tier (Mythos-class, above Opus), lists the current model strings, and reveals how Claude is being positioned across Claude Code, Claude Cowork, and the new browser/spreadsheet/slides agents.
This page covers: what the prompt reveals, what it means for GTM teams running Claude Code workflows, and the full prompt written out with annotations.
What got leaked and why it matters
The leaked file is a system prompt called "Claude Fable 5 — System Prompt". It was posted publicly to GitHub Gist on June 11, 2026.
What it confirms:
- A new model tier exists above Claude Opus. It is called Mythos-class. The two models in it are Claude Fable 5 and Claude Mythos 5.
- Claude Fable 5 and Claude Mythos 5 share the same underlying model. Fable 5 is the generally available version with additional safety measures for dual-use capabilities. Mythos 5 is available without those measures to approved organisations only.
- The current model strings as of June 2026: claude-fable-5, claude-opus-4-8, claude-sonnet-4-6, claude-haiku-4-5-20251001.
- Claude Code and Claude Cowork are named explicitly as the two main agentic products. Claude Code for developers. Claude Cowork for non-developers. Both accessible remotely via the Claude mobile app.
- Three new beta products are named: Claude in Chrome (browsing agent), Claude in Excel (spreadsheet agent), Claude in PowerPoint (slides agent). Claude Cowork can use all three as tools.
What this means for GTM teams
Most GTM teams running Claude Code workflows are currently using claude-sonnet-4-6 or claude-opus-4-8 as their default model. The Mythos-class tier above Opus changes the performance ceiling for complex GTM tasks.
The tasks most likely to benefit from Fable 5 over Sonnet or Opus:
- Multi-step account research that requires synthesising signals across multiple sources
- Complex ICP gap analysis comparing CRM data against enriched market data
- Autonomous campaign launchers that make judgment calls about prospect fit mid-workflow
- MEDDPICC deal qualification that requires deep reasoning about deal health and next actions
- Writing that needs to sound genuinely human across 50+ personalised emails without pattern repetition
The distinction between Fable 5 (safety measures for dual-use) and Mythos 5 (no safety measures, approved orgs only) is unlikely to affect most GTM use cases. Cold email personalisation and ICP research do not trigger the safety measures that differentiate the two models.
The Claude in Excel and Claude in PowerPoint implication
For GTM teams, these two new beta products are the most immediately useful:
Claude in Excel: run enrichment waterfalls, score prospect lists, and build pipeline dashboards without leaving Excel. The spreadsheet agent reads and writes cells like a RevOps analyst.
Claude in PowerPoint: build QBR decks, board presentations, and proposal slides from CRM data without copy-pasting. The slides agent reads structured data and produces formatted slide content.
Claude Cowork using both as tools means a single Cowork session can: pull pipeline data, score it, write a QBR brief, and format it into a slide deck. One session. No manual switching.
The Full Prompt Written Out with GTM Annotations
The full system prompt written out with annotations explaining each section.
Source: gist.github.com/gsans/b3007997f8900003c8ff58125a45e15e
The full prompt
Header and model identity
Product information section
Model strings
Agentic products section
Behaviour rules (refusal handling)
Settings and features
How to use this prompt for GTM
Use it to update your CLAUDE.md model string
Use the memory feature instead of a static CLAUDE.md
Use the style feature for outreach voice
AI Driven, Personalized Outreach on LinkedIn: Prosp 💜