Signal Timeline

x.com2026-06-24trust 6.5SUPPLYUnverified

Claude Code v2.1.190 added strings hinting at Claude Fable 5 returning permanently in subscriptions with weekly limits.

x.com2026-06-24trust 5.0LeakUnverified

Speculation suggests Mythos 5, Anthropic's first 10T reasoning model, is expected later this year.

bloomberg.com2026-06-23trust 8.5SUPPLYMulti-source

An AI startup utilizing Anthropic's latest 'Claude Fable' model has sued the US government over regulatory export restrictions that limit overseas access to the model.

x.com2026-06-23trust 5.0LeakUnverified

According to AndrewCurran_, rumors persist about a new Sonnet model, and Anthropic teams are meeting frequently, with recent discussions possibly having gone well.

bloomberg.com2026-06-22trust 8.5SUPPLYMulti-source

Semiconductor giant Micron has signed a strategic AI infrastructure supply agreement with Anthropic and participated in its funding round, securing critical hardware supply such as HBM (high-bandwidth memory) required for scaling next-generation Claude models.

x.com2026-06-22trust 5.0SUPPLYUnverified

According to AiBattle_, Claude Sonnet 5 is available to select enterprise customers via an Early Access Program as a stop-gap while the next major model's progress has stalled.

x.com2026-06-21trust 5.0LeakUnverified

A leak shared on X indicated that the model slug 'claude-sonnet-5' appeared on an Anthropic partner provider, suggesting an imminent release of the Claude Sonnet 5 model.

x.com2026-06-21trust 5.0LeakUnverified

Claude Sonnet 5 has been spotted as the second release in the Claude 5 series, with reports indicating it is likely to launch on a Tuesday or Thursday.

x.com2026-06-21trust 5.0LeakUnverified

An updated Claude Mythos (potentially Mythos or Fable 5.1) is reportedly on the way, with questions raised about whether it will be made public.

x.com2026-06-21trust 5.0LeakUnverified

A new, more capable version of Mythos has reportedly emerged from training, potentially as Mythos 5.1 or 6, or kept internal.

bloomberg.com2026-06-19trust 8.5SUPPLYMulti-source

While President Trump indicated he does not view Anthropic as a national security threat, it has been reported that select early users continue to maintain preview access to Anthropic's unreleased 'Mythos' model.

x.com2026-06-19trust 5.0SUPPLYUnverified

It was reported that some Project Glasswing users retained access to a preview of Claude Fable 5 despite the broader shutdown.

x.com2026-06-18trust 6.5SUPPLYMulti-source

Anthropic's Managing Director of International stated during a press conference that Claude Fable 5 may become available again in the coming days.

bloomberg.com2026-06-17trust 8.5SUPPLYMulti-source

French President Emmanuel Macron is reportedly consulting with G7 leaders to find workarounds, such as securing alternative deployment routes through trusted partners, to bypass US export restrictions on Anthropic's AI models.

bloomberg.com2026-06-16trust 8.5SUPPLYMulti-source

A warning letter from the US Commerce Secretary to Anthropic revealed that the company was warned to obtain government permission for foreign national access to cutting-edge AI models, which led Anthropic to disable certain models like Mythos due to export regulations.

bloomberg.com2026-06-15trust 8.5SUPPLYMulti-source

Following a U.S. government order, Anthropic has blocked international access to its latest models, and company executives are set to meet with government officials to discuss the restriction.

x.com2026-06-15trust 5.0SUPPLYMulti-source

Information regarding Claude Fable 5's export-control context has been discussed.

theverge.com2026-06-14trust 7.0SUPPLYMulti-source

The U.S. government has ordered Anthropic to block overseas access to "Fable 5" and the unreleased model "Mythos 5" due to national security concerns.

anthropic.com2026-06-12trust 10.0Official hintOfficial

Anthropic announced a multi-year global partnership with DXC Technology to integrate Claude into regulated industry systems, signaling strong confidence in the enterprise deployment and stability of future Claude models, including claude-next.

bloomberg.com2026-06-11trust 8.5Official hintMulti-source

OpenAI is considering significant price cuts in anticipation of Anthropic's predicted price reductions, suggesting Anthropic may intensify competition with new models like claude-next.

x.com2026-06-10trust 6.5LeakUnverified

testingcatalog highlighted on June 10 that Voice Mode on Claude will soon include a model selector (currently showing Claude Haiku 4.5), signaling a major upgrade and possible non-TTS voice support tied to the new model family.

anthropic.com2026-06-09trust 10.0Model IDOfficial

Anthropic officially launched "Claude Fable 5" on June 9, 2026. This model is generally available, and its API model ID is `claude-fable-5`.

anthropic.com2026-06-09trust 10.0Official hintOfficial

Anthropic officially released its most capable generally available AI, "Claude Fable 5", and "Claude Mythos 5", the same underlying model with some safeguards lifted for vetted partners in Project Glasswing.

openrouter.ai2026-06-09trust 9.0Model IDOfficial

New model ID "anthropic/claude-fable-5" (Anthropic: Claude Fable 5) appeared in the OpenRouter model list.

vertexaisearch.cloud.google.com2026-06-09trust 8.0SUPPLYMulti-source

On June 9, 2026, Anthropic introduced Claude Fable 5 and Mythos 5 with a substantial price reduction, offering them at $10 per million input tokens and $50 per million output tokens, which is a 60% decrease from the previous Claude Mythos Preview rates.

x.com2026-06-09trust 6.5Model IDUnverified

On June 9, @chetaslua shared new "Claude Fruitcake" checkpoints were added for testing, stating it was "possibly launching as soon as tomorrow."

x.com2026-06-09trust 6.5LeakMulti-source

testingcatalog reported on June 9 that Claude Fable 5 (Mythos) began rolling out on Claude.ai and APIs, with a usage plan active until June 22 (2x Opus usage); also noted preparation for Google Cloud Platform deployment 'really soon'.

x.com2026-06-09trust 6.5LeakMulti-source

Claude Mythos is reportedly planned for release as 'Claude Fable 5' according to checkpoints, Dev Mode, and Hacker News sources; Anthropic is also hosting a developer event in Japan tomorrow.

x.com2026-06-09trust 6.5LeakMulti-source

Claude Fable 5 is reportedly available in a usage plan until June 22, equivalent to twice the Opus usage, with pricing rumored to be twice that of Opus 4.8 according to The Information.

x.com2026-06-09trust 5.0Model IDMulti-source

On June 9, @M1Astra and @koltregaskes posted that new "Claude Fruitcake EAP" checkpoints were detected over the weekend.

x.com2026-06-09trust 5.0LeakMulti-source

Anthropic is reportedly 'preparing to release Claude Mythos 5 tomorrow', expected to be the first 10T+ parameter model.

x.com2026-06-09trust 5.0LeakMulti-source

Anthropic is reportedly releasing 'Claude Fable 5 today', described as the public and safer version of Mythos 5, establishing a new Mythos-class tier above Opus.

x.com2026-06-09trust 5.0LeakMulti-source

The new Anthropic model, 'Claude Fable' (the public version of Mythos), is expected to be released today, earlier than anticipated.

reddit.com2026-06-07trust 4.5LeakUnverified

A model string named "Claude Oceanus V1" was spotted in Anthropic's console, suggesting it could be related to future Mythos-class AI agent work.

x.com2026-06-06trust 6.5LeakMulti-source

A new 'Claude Mythos 5' model slug was spotted in Dev Mode as its own class, separate from Haiku/Sonnet/Opus, suggesting an imminent release.

x.com2026-06-06trust 6.5LeakMulti-source

'Claude-Mythos-5' was spotted, indicating Mythos 5 would be released first, followed by Opus 5/Sonnet 5.

x.com2026-06-06trust 5.0LeakUnverified

On June 6, AILeaksAndNews shared early outputs from the upcoming Mythos model (generating Minecraft/Terraria-style worlds).

x.com2026-06-06trust 5.0LeakMulti-source

Claude Mythos 5 was identified as the first 14-digit-parameter model, with a release 'likely this week'.

openrouter.ai2026-06-03trust 9.0Model IDUnverified

Around June 3, 2026, the model identifier `claude-oceanus-v1-p` was sighted within the Anthropic Console and through unauthorized API proxy services, reportedly associated with a "Mythos" family model undergoing red-team testing prior to its formal release.

reddit.com2026-06-03trust 4.5LeakMulti-source

An unreleased Anthropic model ID, "claude-oceanus-v1-p," was reportedly observed in the Claude Console Workbench for red teamers on June 3, 2026. This model is believed to be a further development in the Mythos line. API access to this model was allegedly resold on a black market, leading Anthropic to temporarily halt the red-team program and initiate an internal investigation.

reddit.com2026-06-03trust 4.5Model IDUnverified

Around June 3, 2026, a new Anthropic model identifier, 'claude-oceanus-v1-p', was spotted in the Claude Console Workbench and circulated among researchers, reportedly made available to red teamers for security testing. Reports suggest this 'Oceanus' model builds on the Claude Mythos Preview foundation, indicating it is a part of the Mythos lineage.

fastcompany.com2026-06-01trust 6.0LeakMulti-source

Credible journalist reports and leaks indicate that Anthropic's next model, "Oceanus" (model string `claude-oceanus-v1-p`), has been spotted in the Anthropic Console Workbench and made available to red teams, suggesting a potential wider release soon.

anthropic.com2026-05-28trust 10.0Official hintOfficial

Anthropic announced the release of Claude Opus 4.8 and simultaneously hinted at plans to release a new class of models with higher intelligence than Opus, referring to the Mythos-class models currently used by a small number of organizations through Project Glasswing.

anthropic.com2026-05-28trust 10.0Official hintOfficial

Anthropic updated its Opus model to Claude Opus 4.8 on May 28, 2026. This update includes benchmark improvements and new features.