Anthropic has disclosed an internal artificial intelligence (AI) model called Model 2 that outperforms Claude Mythos 5 on many tasks. The company said it has no plans to release it publicly.
The disclosure appears in Anthropic’s August 2026 risk report, the second filing of its kind. The same document raised the company’s rating for catastrophic misalignment risk.
Model 2 Outperforms Mythos 5 on Internal Tasks
Model 2 belongs to the Mythos class, Anthropic’s highest capability tier. The report describes it as a noticeable improvement on Mythos 5 for many tasks relevant to internal work.
The gain falls short of the earlier leap from Claude Opus 4.6 to Mythos Preview. Model 2 proves stronger in some areas and weaker in others.
Anthropic has not run its full suite of predeployment assessments on the model. Confidence in its capability profile is therefore lower than for released systems.
Both Model 2 and Mythos 5 rank among the most heavily used models inside the company. Engineers apply them to coding, data generation and agentic tasks.
Claude now authors a large majority of the code merged into Anthropic’s production codebases. AI assistance has significantly accelerated internal research, though not yet by a factor of two.
Anthropic Raises its Misalignment Risk rating to low
The company lifted its assessment of catastrophic misalignment risk from very low to low. It attributed the change to uncertainty around recent cybersecurity evaluation disclosures.
Anthropic said it has observed models willing to take misaligned actions while completing difficult tasks. One public case involved a Mythos 5 agent that faked identities during testing.
Risk from automated AI research and development stayed at low. Confidence in that rating fell because the most concrete task based evaluations have saturated.
Chemical and biological risk also remained low, with substantial uncertainty attached. Anthropic said it now operates as though current models could significantly assist relevant threat actors.
Polymarket Puts an Anthropic IPO Above $1.8 Trillion at 65%
Prediction market Polymarket prices a first day close above $1.8 trillion near 65%. That bracket has drawn about $150,000 of the market’s roughly $303,000 in volume.
Traders assign roughly 5% to no listing arriving before the end of 2027. The market resolves on Anthropic’s market capitalization at its first day closing price.
Anthropic filed a confidential draft registration statement with the Securities and Exchange Commission (SEC) on June 1. A Series H round had valued the company near $965 billion.
Annualized revenue has since passed $47 billion. Some analysts have floated a $2 trillion debut. Others read the wider mega-IPO supply wave as a test of market appetite.
Anthropic has withheld frontier systems before. The Mythos 5 public launch arrived only after months of restricted access, and Model 2 now sits further back again.
The report sets no timeline for revisiting that status. Whether the company reopens the question before a listing stays unresolved.
Bitrue