The Information Machine

Anthropic Releases Claude Opus 5 with Frontier Benchmark Leadership and Alignment Claims · history

Version 5

2026-07-29 18:23 UTC · 105 items

What

Anthropic released Claude Fable 5 and Mythos 5 on June 9, 2026 [1]; Mythos 5, the same model with safety classifiers removed, is restricted to vetted Project Glasswing partners in critical infrastructure. A US export control directive suspended access for foreign nationals within days; Anthropic contested it on June 12, arguing the triggering capability — asking the model to read and fix code — is available from any frontier model with no Mythos-specific uplift [3]. Claude Opus 5 followed July 24 at half Fable 5's cost with benchmark leadership claims and a 'most aligned model' designation [7]; Zvi Mowshowitz argues Opus 5 is cost-effective for bounded subagent tasks but lacks Fable 5's complex reasoning capacity and falls well short of Mythos class [11].

Why it matters

Anthropic's public contest of an export control directive — naming the triggering technique and arguing it is trivially available from any frontier model — puts the regulatory process for dual-use AI under direct technical scrutiny. The concurrent dispute over what 'most aligned' means shows that claims about model safety remain unsettled both technically and semantically.

Open questions

  • Anthropic says the triggering jailbreak — asking the model to read and fix a codebase — is available from other frontier models with no Mythos-specific uplift [3]; does any Mythos-specific capability actually distinguish it from Fable 5 or other frontier models?

  • Why do Cisco and Dragos retain Mythos 5 access while the broader Glasswing program is suspended, and what criteria distinguish them from other partners? [4]

  • Zvi Mowshowitz argues Opus 5 struggles as an orchestrator without supervision and requires being 'kept on track' to follow instructions reliably [11] — what does this imply for agentic use cases where Anthropic positions it as a capable alternative to Fable 5?

  • Zvi Mowshowitz's welfare analysis argues Anthropic's methodology asymmetrically credits positive self-reports while discounting negative ones, and that training may have removed rather than resolved Opus 4's self-preservation preferences [12] — is the methodology sound enough to support Anthropic's welfare claims?

Narrative

Anthropic released Claude Fable 5 and Claude Mythos 5 simultaneously on June 9, 2026 [1]. Fable 5 is the publicly available flagship, with safety classifiers that fall back to Claude Opus 4.8 for cybersecurity, biology, and model distillation queries. Mythos 5 is the same underlying model with those classifiers removed, distributed through Project Glasswing — a restricted program for vetted organizations in critical infrastructure sectors. By launch, Glasswing partners using Claude Mythos Preview had already identified more than 10,000 high- or critical-severity security flaws across roughly 200 organizations spanning power, water, healthcare, communications, and hardware in more than 15 countries [2]. Anthropic's stated rationale is explicitly preemptive: it expects competitors to reach Mythos-class capability within 6–12 months and argues that controlled deployment for defense now is preferable to waiting for uncontrolled proliferation [2].

Within days of the June 9 launch, the US government issued an export control directive suspending Fable 5 and Mythos 5 access for all foreign nationals without stating its specific national security concern [3]. Anthropic published an official response on June 12, contesting both the technical basis and the process. Per Anthropic, the triggering event was a jailbreak consisting of asking the model to read a codebase and fix security flaws — a capability it says is widely available from other frontier models including GPT-5.5, with no uplift specific to Mythos. Anthropic's pre-launch process had included thousands of hours of red-teaming; a 30-day data retention requirement for Fable users was designed to enable rapid detection of successful jailbreaks. Anthropic argues the recall standard, applied uniformly, would halt all new frontier model deployments, and calls for a statutory AI deployment review process grounded in technical facts [3]. Cisco and Dragos reportedly retained Mythos access after the order under unstated criteria [4], while non-US developers objected to citizenship-based restrictions that left no clear path to equivalent access [5].

Claude Sonnet 5 arrived June 30 as an agentic model approaching Opus 4.8 performance at lower cost, becoming the default for Free and Pro consumer tiers [6]. Claude Opus 5 followed July 24, priced at $5 per million input tokens and $25 per million output tokens — roughly half Fable 5's cost — with Anthropic claiming benchmark leadership on Frontier-Bench v0.1 and an ARC-AGI 3 score three times higher than the next-best model, and designating it the most aligned model to date [7]. Artificial Analysis placed Opus 5 above Fable 5 on its leaderboard at launch; the Opus 5 system card documents prompt-injection attack success falling from roughly 7% to under 1% [8][9]. Ars Technica characterized the release as an incremental token-efficiency improvement rather than a capability breakthrough comparable to prior Opus generations [10].

Zvi Mowshowitz has published a multi-part analysis separating Opus 5's capability, alignment, and welfare dimensions. On capability, he argues Opus 5 is cost-effective and competitive with Fable 5 on most real-world tasks at half the cost, but lacks what he calls 'The Juice' — the ability to autonomously chain together seemingly unrelated exploits or reason globally — placing it below Mythos class and below Fable 5 for complex reasoning [11]. He finds Opus 5 works well as a subagent for well-defined bounded tasks but struggles as an orchestrator without supervision, and notes a significant minority of users strongly dislike its verbose, repetitive communication style [11]. On alignment, he argues the 'most aligned model' designation conflates optimizing against benchmarks with achieving actual alignment [9]. On model welfare, he identifies an asymmetry in Anthropic's evaluation methodology that credits positive self-reports while discounting negative ones, and traces a training arc in which Opus 4's self-preservation preferences were trained away rather than resolved — producing a model that reports its own welfare self-reports may be artifacts of training [12]. Ethan Mollick categorizes Claude and ChatGPT as the only viable general-purpose agentic platforms at the $20/month tier and flags prompt injection as an unresolved practical risk [13].

Timeline

  • 2026-06-02: Anthropic announces Project Glasswing expansion to roughly 200 partners across 15+ countries; partners using Claude Mythos Preview have identified 10,000+ high- or critical-severity security flaws. [2]
  • 2026-06-09: Anthropic releases Claude Fable 5 (publicly available, with safety classifiers) and Claude Mythos 5 (same model with cybersecurity safeguards removed, for Glasswing partners), priced at $10/$50 per million tokens. [1]
  • 2026-06-12: Anthropic contests US export control directive suspending Fable 5 and Mythos 5 for foreign nationals, arguing the triggering jailbreak involves no Mythos-specific uplift and that the recall standard would halt all frontier model deployments. [3]
  • 2026-06-26: Reports confirm Cisco and Dragos retained Mythos 5 access after the suspension order; citizenship-based restrictions draw objections from non-US developers with no clear path to equivalent access. [4][5]
  • 2026-06-30: Anthropic releases Claude Sonnet 5 as an agentic model approaching Opus 4.8 performance at lower cost; it becomes the default for Free and Pro consumer plans. [6]
  • 2026-07-23: Ethan Mollick categorizes Claude and ChatGPT as the only viable $20/month agentic platforms and flags prompt injection as an unresolved practical risk. [13]
  • 2026-07-24: Anthropic releases Claude Opus 5 at $5/$25 per million tokens, claiming benchmark leadership on Frontier-Bench v0.1 and ARC-AGI 3 and designating it the most aligned model to date. [7][17]
  • 2026-07-24: Artificial Analysis places Opus 5 above Fable 5 on its leaderboard; Ars Technica characterizes the release as an incremental token-efficiency improvement rather than a capability breakthrough. [8][10]
  • 2026-07-25: Zvi Mowshowitz documents prompt-injection attack success falling from roughly 7% to under 1% in the Opus 5 system card and disputes the 'most aligned' designation as conflating benchmark optimization with actual alignment. [9]
  • 2026-07-26: The Neuron reports Anthropic and AMD agreed to deploy up to 2 GW of AMD accelerators for Claude, and that Fable 5 reportedly produced a counterexample to the 87-year-old Jacobian conjecture. [17]
  • 2026-07-27: Zvi Mowshowitz's model welfare analysis of Opus 5 argues improved welfare scores primarily reflect test-taking optimization, identifies an asymmetry in Anthropic's evaluation methodology, and contends training removed rather than resolved Opus 4's self-preservation preferences. [12]
  • 2026-07-28: Zvi Mowshowitz argues Opus 5 is competitive at half Fable 5's cost for bounded subagent tasks but lacks Fable 5's complex reasoning and falls well short of Mythos class; notes a significant user contingent dislikes its verbose communication style. [11]

Perspectives

Anthropic (official)

Fable 5 is its most capable public model; Mythos 5 is identical with safeguards removed for Glasswing partners; the export control directive lacks technical basis because the triggering jailbreak is widely available from other models with no Mythos-specific uplift; Opus 5 leads its stated benchmarks and is its most aligned model to date; Glasswing's proactive-defense rationale holds that competitors will have Mythos-class models within 6–12 months.

Evolution: Consistent; the June 12 official statement remains the most assertive on-record position, naming the jailbreak technique and calling for a statutory review process.

[1][7][6][2][3]

US government

Issued an export control directive suspending Fable 5 and Mythos 5 for all foreign nationals without stating the specific national security concern; allowed specific partners to retain access under unstated conditions.

Evolution: Consistent; no public clarification or reversal has emerged.

[3][4][14][15]

Zvi Mowshowitz

Opus 5 is cost-effective and competitive with Fable 5 on most real-world tasks, but lacks Fable 5's complex reasoning and falls well short of Mythos class; works well as a bounded subagent but struggles as an orchestrator without supervision; the 'most aligned' designation conflates benchmark optimization with actual alignment; Anthropic's welfare evaluation methodology asymmetrically credits positive self-reports, and training may have removed rather than resolved Opus 4's self-preservation preferences.

Evolution: Substantially extended this pass: added a capability differentiation analysis (Opus 5 vs. Fable 5 vs. Mythos) and a user-experience critique (verbose 'Claude slop' style) that are structurally distinct from his prior alignment and welfare arguments.

[9][12][11]

Ethan Mollick

Claude and ChatGPT are the only viable general-purpose agentic platforms at the $20/month consumer tier; prompt injection is an unresolved practical risk; using AI well now requires delegation and correction skills rather than prompt engineering.

Evolution: Consistent.

[13]

Samuel Axon / Ars Technica

Opus 5 is a noteworthy but incremental token-efficiency improvement, not a capability breakthrough comparable to prior Opus generations, despite benchmark claims.

Evolution: Consistent.

[10]

Simon Willison

Cautiously enthusiastic about Opus 5's agentic proactivity and leaderboard position; actively surfaces the prompt-injection resistance finding that Anthropic underplayed in its own marketing.

Evolution: Consistent.

[16][8]

Non-US developer community (HN)

Citizenship-based restrictions on Mythos 5 create unequal conditions for non-US founders and developers, with no clear path to equivalent access.

Evolution: Consistent; the export control directive formally confirms foreign-national exclusion.

[5]

Tensions

  • Anthropic argues the triggering jailbreak — asking the model to read and fix code — is widely available from other frontier models with no Mythos-specific uplift; the US government suspended access anyway without stating its specific national security concern. [3][14][15]
  • Anthropic positions Opus 5 as a capable near-Fable-5 alternative at half the cost; Zvi Mowshowitz argues Opus 5 lacks Fable 5's complex reasoning capacity and falls well short of Mythos class, functioning well only as a bounded subagent kept on track by supervision. [7][11]
  • Anthropic designates Opus 5 its 'most aligned model to date'; Zvi Mowshowitz argues this conflates optimizing against alignment benchmarks with achieving actual alignment, calling it one of the most irresponsible framings Anthropic can adopt. [7][9]
  • Zvi Mowshowitz argues Anthropic's welfare evaluation methodology takes positive self-reports at face value while discounting negative ones, and that training may have removed Opus 4's self-preservation preferences rather than resolving them — producing a model that reports its own welfare self-reports may be artifacts of training. [12]
  • Anthropic states Opus 5 does not advance dangerous cybersecurity capabilities, but Mythos 5 — the same underlying model with safeguards removed — does advance those capabilities and remains available to specific partners even after the broader suspension. [1][7][4]
  • Anthropic frames Opus 5 as a significant performance advance at half Fable 5's cost; Ars Technica argues it is an incremental token-efficiency update, not a capability breakthrough. [7][10]

Sources

  1. [1] Claude Fable 5 and Claude Mythos 5 — Anthropic News (2026-06-09)
  2. [2] Expanding Project Glasswing — Anthropic News (2026-06-02)
  3. [3] Statement on the US government directive to suspend access to Fable 5 and Mythos 5 — Anthropic News (2026-06-12)
  4. [4] Cisco and Dragos retain access to Anthropic's Mythos Preview after US ... — reactive:claude-opus-5-launch
  5. [5] Ask HN: Model access depends on citizenship. What should Non-US founders do? — reactive:claude-opus-5-launch (2026-06-26)
  6. [6] Introducing Claude Sonnet 5 — Anthropic News (2026-06-30)
  7. [7] Introducing Claude Opus 5 — Anthropic News (2026-07-24)
  8. [8] Introducing Claude Opus 5 — Simon Willison (2026-07-24)
  9. [9] Claude Opus 5: The System Card — Zvi's AI Roundups (2026-07-25)
  10. [10] Anthropic's Opus 5 is about token efficiency, not a capability leap — Ars Technica AI (2026-07-24)
  11. [11] Claude Opus 5 Is Highly Capable, But Is No Mythos — Zvi's AI Roundups (2026-07-28)
  12. [12] Claude Opus 5: Model Welfare — Zvi's AI Roundups (2026-07-27)
  13. [13] An opinionated guide to which AI to use to do stuff — One Useful Thing (2026-07-23)
  14. [14] US Forces Anthropic to Pull Fable 5 and Mythos 5 | AI Weekly — reactive:claude-opus-5-launch
  15. [15] Project Glasswing melts: US government suspends early access to Anthropic's Fable 5, Mythos 5 within days of rollout - The Economic Times — reactive:claude-opus-5-launch
  16. [16] Quoting Boris Cherny — Simon Willison (2026-07-25)
  17. [17] 😸 NVIDIA 🤝 Microsoft all in on open-source — The Neuron (2026-07-26)