● Covered by 14 sources · 31 reportsMedium impact1 negative29 neutral1 positive
Claude Mythos 5 AI Cybersecurity Capabilities Expanded, $35M Fund for Open-Source Security
Key points
- Claude Mythos 5 is now in Claude Security and integrating with partner tools.
- A $35 million Defender Advantage Fund launched for open-source security.
- The Cyber Verification Program will be expanded.
- These efforts aim to expand defensive AI access with safety measures.
- Project Glasswing launched in April.
- Project Glasswing gave early access to Claude Mythos Preview and Mythos 5.
- Claude Fable 5 is a broadly available model that blocks dual-use cyber work.
- Claude Fable 5.1 and Claude Mythos 5.1 are new AI models.
- Claude Fable 5.1 and Claude Mythos 5.1 are designed for coding and knowledge work.
- Claude Fable 5.1 is generally available.
- Claude Mythos 5.1 is for trusted access programs.
- Mythos 5.1 safeguards support cybersecurity and life sciences.
- Fable 5.1 costs 25% less than Fable 5 for typical workloads.
- Fable 5.1 costs up to 45% less for highly agentic work.
- Enterprise Frontier Safeguards (EFS) provide complete data privacy.
- Fable 5.1 was released three months after Fable 5.
- Fable 5.1 sets a new standard on coding, knowledge work, and long-running problem-solving tasks.
- Fable 5.1 found the cause of a rare crash for Millennium.
- Fable 5.1 and Mythos 5.1 are the same underlying model.
- Fable 5.1 has a 75% cost reduction for cached context reads.
- EFS allows organizations to retain monitoring data inside infrastructure they control.
- Fable 5.1 is available on Amazon Bedrock and Claude Platform on AWS.
- Fable 5.1 is a Covered Model with additional data retention, safety review, and access policies.
- Fable 5.1 pricing is $0.25 per million for cache reads.
- Fable 5.1 is restricted to API usage and users with Max, Team Premium, and Enterprise Premium plans.
- Pro subscribers can use Fable 5.1 via usage credits.
- Fable 5.1 outperforms Opus 5 and OpenAI’s GPT-5.6 Sol in benchmarks.
- EFS will roll out to users in the fall.
- Anthropic has never trained on enterprise data.
- Fable 5.1 addresses customer feedback on cost, data retention, and overzealous safeguards.
- Dan Shipper of Every CEO stated Fable 5.1 is the strongest, fast, and token-efficient coding model.
- Box CEO Aaron Levie stated Fable 5.1 picked up subtleties Fable 5 missed.
- Lisan al Gaib noted Mythos 5.1 with low reasoning scores the same as its predecessor set to Max reasoning.
- Min Choi collected examples of people one-shotting games, building 3D worlds, and creating insane simulations within 24 hours of Fable 5.1 release.
- Fable 5.1 scores 52.6% on the Terminal-Bench-Science benchmark.
- Fable 5 scores 24.7% on the Terminal-Bench-Science benchmark.
- Terminal-Bench-Science benchmark gives a model a terminal and multi-step scientific research tasks.
- Fable 5.1 costs $10 per million input tokens.
- Fable 5.1 costs $50 per million output tokens.
- Fable 5.1 is built for long-running, high-stakes work that runs for hours and spans many applications.
- Fable 5.1 can own more of a software project on its own, handling features across an entire codebase, code review, and performance work over extended sessions.
- Fable 5.1 is subject to data retention for up to 30 days and human review by Amazon personnel.
- A new aws_review data retention mode retains prompts and outputs for human safety review within the AWS boundary.
- The provider_data_share mode is legacy.
- Amazon Bedrock does not share data with the model provider.
- Anthropic's Fable 5.1 benchmark score was produced under conditions most users don't have access to.
- The benchmark allows each model up to eight hours per task.
- The benchmark's leaderboard tested Fable 5 through Claude Code at maximum effort.
- Testing Fable 5 on the benchmark cost $14,180 across 210 attempts, about $67 each.
- Anthropic released Claude Opus 5.5.
- OpenAI launched GPT-6 Sol and GPT-6 Luna.
- Claude Opus 5.5 is better at handling complex work, finding and fixing software inefficiencies, financial analysis, and business work.
- Claude Opus 5.5 scored better at agentic coding than GPT-6 Astra on Terminal-Bench 4.0 and FrontierCode v1.1 (Main).
- Claude Opus 5.5 produces writing that is easier to understand.
- Claude Opus 5.5 is the first release in the Claude 5.5 family.
- Claude Opus 5.5 costs 40% less to run than Opus 5.
- Mario Rodriguez, GitHub chief product officer, stated Opus 5.5 used among the fewest tokens and steps in testing.
- Opus 5.5 solved more terminal tasks than Opus 5 in less than half the steps in VS Code.
- Claude Opus 5.5 requires less compute to serve than Opus 5.
- Opus 5.5 input tokens cost $4 per million.
- Opus 5.5 output tokens cost $20 per million.
- Opus 5.5 cache reads cost $0.20 per million tokens.
- Opus 5.5 generates output more than 30% faster than Opus 5.
- Opus 5.5 beats Fable 5.1 in every metric.
- Opus 5.5 beats OpenAI’s GPT-6 Astra in virtually every category.
- Opus 5.5 is available on all Claude platforms.
- Opus 5.5 has four breaking changes that can cause existing agents to fail.
- Opus 5.5 returns a 400 error if thinking is disabled or uses enabled with budget_tokens.
- Claude Opus 5.5 has zero data retention available.
- Opus 5.5 costs $4 per million input tokens, down from $5 for Opus 5.
- Opus 5.5 costs $20 per million output tokens, down from $25 for Opus 5.
- The price cut for Opus 5.5 accounts for a 20% saving.
- Opus 5.5 did not outperform Opus 5 on complex reasoning tasks.
- Claude Sonnet 5.5 is the second model in the Claude 5.5 family.
- A new version of Claude Haiku will launch in the coming weeks.
- Sonnet 5.5 generates output more than 30% faster than Sonnet 5.
- Sonnet 5.5 lowers the cost per task by up to 30% compared to Sonnet 5.
- Sonnet 5.5 is the company's fastest Sonnet model yet.
- Sonnet 5.5 is strongest at well-scoped everyday tasks, fixing bugs, and creating polished documents, slides, and spreadsheets.
- Sonnet 5.5 adds polish to user interfaces and can follow slide templates.
- Sonnet 5.5 has significant cyber capabilities, comparable to Opus 5.
- Sonnet 5.5 performs better than Opus 5.5 on agentic coding.
- Sonnet 5.5 is for the cost-conscious customer who needs less intelligence.
- Theo Chu is a research product manager at Anthropic.
- Sonnet 5.5 is the best medium-sized model.
- Sonnet 5.5 keeps Sonnet 5's pricing: $2 per million input tokens, $10 per million output tokens.
- Sonnet 5.5 has cyber safeguards and model fallbacks.
- Sonnet 5.5 scores 70.6% on Terminal-Bench 4.0.
- Opus 5.5 scores 66.4% on Terminal-Bench 4.0 at xhigh effort setting.
- Opus 5.5 was released on September 22.
- Opus 5.5 leads Fable 5.1 on Terminal-Bench 4.0, FrontierCode, and CursorBench.
- Anthropic's model docs still recommend Fable 5.1 for demanding reasoning and long-horizon agentic work.
- Anthropic's model docs list Fable 5.1 as the slower model.
- Opus 5.5 matched Opus 5 on reasoning problems at a lower cost in a previous test.
- Opus 5.5's speed gain fell short of Anthropic's claim in a previous test.
- Livenerf is a new benchmark to track Claude Opus 5.5 performance over time.
- Livenerf detects degradation in large language models post-launch.
- Livenerf provides a day-0 baseline for model consistency.
- Livenerf runs on a Claude Max subscription through headless Claude Code.
- Livenerf uses frozen prompts, pinned CLI, exact graders, and raw logs.
- Livenerf measures drift statistically over thousands of samples.
- Livenerf is built on Inspect.
Expanded Access to Claude Mythos 5
Claude Mythos 5, an advanced AI model for cybersecurity, is now integrated into Claude Security for enterprise customers. This allows users to scan codebases for vulnerabilities and receive patch suggestions. The model is also being integrated into cybersecurity tools from various partners, making its capabilities accessible through existing defensive platforms.
Safeguarding AI for Defensive Use
The expansion of Claude Mythos 5 follows Project Glasswing, which initially provided the model to a select group of organizations to identify and fix vulnerabilities. The company has developed safety classifiers and safeguards to allow broader access to Mythos-class models, focusing on providing defensive outputs while restricting direct access that could lead to malicious use.
$35 Million Fund for Open-Source Security
A new initiative, the Defender Advantage Fund (0xDAF), has been launched with $35 million in credits. This fund will support organizations working to patch vulnerabilities in open-source projects, automate scanning and patching processes for open-source software, and explore new security approaches. This aims to bolster the security of widely used open-source components.
Cyber Verification Program Expansion
The company also announced plans to expand its Cyber Verification Program. This program currently offers vetted defenders reduced safeguards on Opus, indicating a continued effort to provide advanced AI capabilities to trusted entities for cybersecurity purposes.
Updates
- Livenerf is a new benchmark to track Claude Opus 5.5 performance over time.
- Livenerf detects degradation in large language models post-launch.
- Livenerf provides a day-0 baseline for model consistency.
- Livenerf runs on a Claude Max subscription through headless Claude Code.
- Livenerf uses frozen prompts, pinned CLI, exact graders, and raw logs.
- Livenerf measures drift statistically over thousands of samples.
- Livenerf is built on Inspect.
- Opus 5.5 was released on September 22.
- Opus 5.5 leads Fable 5.1 on Terminal-Bench 4.0, FrontierCode, and CursorBench.
- Anthropic's model docs still recommend Fable 5.1 for demanding reasoning and long-horizon agentic work.
- Anthropic's model docs list Fable 5.1 as the slower model.
- Opus 5.5 matched Opus 5 on reasoning problems at a lower cost in a previous test.
- Opus 5.5's speed gain fell short of Anthropic's claim in a previous test.
- Sonnet 5.5 has cyber safeguards and model fallbacks.
- Sonnet 5.5 scores 70.6% on Terminal-Bench 4.0.
- Opus 5.5 scores 66.4% on Terminal-Bench 4.0 at xhigh effort setting.
- Claude Sonnet 5.5 is the second model in the Claude 5.5 family.
- A new version of Claude Haiku will launch in the coming weeks.
- Sonnet 5.5 generates output more than 30% faster than Sonnet 5.
- Sonnet 5.5 lowers the cost per task by up to 30% compared to Sonnet 5.
- Sonnet 5.5 is the company's fastest Sonnet model yet.
- Sonnet 5.5 is strongest at well-scoped everyday tasks, fixing bugs, and creating polished documents, slides, and spreadsheets.
- Sonnet 5.5 adds polish to user interfaces and can follow slide templates.
- Sonnet 5.5 has significant cyber capabilities, comparable to Opus 5.
- Sonnet 5.5 performs better than Opus 5.5 on agentic coding.
- Sonnet 5.5 is for the cost-conscious customer who needs less intelligence.
- Theo Chu is a research product manager at Anthropic.
- Sonnet 5.5 is the best medium-sized model.
- Sonnet 5.5 keeps Sonnet 5's pricing: $2 per million input tokens, $10 per million output tokens.
- Opus 5.5 costs $4 per million input tokens, down from $5 for Opus 5.
- Opus 5.5 costs $20 per million output tokens, down from $25 for Opus 5.
- The price cut for Opus 5.5 accounts for a 20% saving.
- Opus 5.5 did not outperform Opus 5 on complex reasoning tasks.
- Claude Opus 5.5 has zero data retention available.
- Opus 5.5 has four breaking changes that can cause existing agents to fail.
- Opus 5.5 returns a 400 error if thinking is disabled or uses enabled with budget_tokens.
- Anthropic released Claude Opus 5.5.
- OpenAI launched GPT-6 Sol and GPT-6 Luna.
- Claude Opus 5.5 is better at handling complex work, finding and fixing software inefficiencies, financial analysis, and business work.
- Claude Opus 5.5 scored better at agentic coding than GPT-6 Astra on Terminal-Bench 4.0 and FrontierCode v1.1 (Main).
- Claude Opus 5.5 produces writing that is easier to understand.
- Claude Opus 5.5 is the first release in the Claude 5.5 family.
- Claude Opus 5.5 costs 40% less to run than Opus 5.
- Mario Rodriguez, GitHub chief product officer, stated Opus 5.5 used among the fewest tokens and steps in testing.
- Opus 5.5 solved more terminal tasks than Opus 5 in less than half the steps in VS Code.
- Claude Opus 5.5 requires less compute to serve than Opus 5.
- Opus 5.5 input tokens cost $4 per million.
- Opus 5.5 output tokens cost $20 per million.
- Opus 5.5 cache reads cost $0.20 per million tokens.
- Opus 5.5 generates output more than 30% faster than Opus 5.
- Opus 5.5 beats Fable 5.1 in every metric.
- Opus 5.5 beats OpenAI’s GPT-6 Astra in virtually every category.
- Opus 5.5 is available on all Claude platforms.
- Anthropic's Fable 5.1 benchmark score was produced under conditions most users don't have access to.
- The benchmark allows each model up to eight hours per task.
- The benchmark's leaderboard tested Fable 5 through Claude Code at maximum effort.
- Testing Fable 5 on the benchmark cost $14,180 across 210 attempts, about $67 each.
- Fable 5.1 is built for long-running, high-stakes work that runs for hours and spans many applications.
- Fable 5.1 can own more of a software project on its own, handling features across an entire codebase, code review, and performance work over extended sessions.
- Fable 5.1 is subject to data retention for up to 30 days and human review by Amazon personnel.
- A new aws_review data retention mode retains prompts and outputs for human safety review within the AWS boundary.
- The provider_data_share mode is legacy.
- Amazon Bedrock does not share data with the model provider.
- Min Choi collected examples of people one-shotting games, building 3D worlds, and creating insane simulations within 24 hours of Fable 5.1 release.
- Fable 5.1 scores 52.6% on the Terminal-Bench-Science benchmark.
- Fable 5 scores 24.7% on the Terminal-Bench-Science benchmark.
- Terminal-Bench-Science benchmark gives a model a terminal and multi-step scientific research tasks.
- Fable 5.1 costs $10 per million input tokens.
- Fable 5.1 costs $50 per million output tokens.
- Fable 5.1 addresses customer feedback on cost, data retention, and overzealous safeguards.
- Dan Shipper of Every CEO stated Fable 5.1 is the strongest, fast, and token-efficient coding model.
- Box CEO Aaron Levie stated Fable 5.1 picked up subtleties Fable 5 missed.
- Lisan al Gaib noted Mythos 5.1 with low reasoning scores the same as its predecessor set to Max reasoning.
- Fable 5.1 and Mythos 5.1 are the same underlying model.
- Fable 5.1 has a 75% cost reduction for cached context reads.
- EFS allows organizations to retain monitoring data inside infrastructure they control.
- Fable 5.1 is available on Amazon Bedrock and Claude Platform on AWS.
- Fable 5.1 is a Covered Model with additional data retention, safety review, and access policies.
- Fable 5.1 pricing is $0.25 per million for cache reads.
- Fable 5.1 is restricted to API usage and users with Max, Team Premium, and Enterprise Premium plans.
- Pro subscribers can use Fable 5.1 via usage credits.
- Fable 5.1 outperforms Opus 5 and OpenAI’s GPT-5.6 Sol in benchmarks.
- EFS will roll out to users in the fall.
- Anthropic has never trained on enterprise data.
- Claude Fable 5.1 and Claude Mythos 5.1 are new AI models.
- Claude Fable 5.1 and Claude Mythos 5.1 are designed for coding and knowledge work.
- Claude Fable 5.1 is generally available.
- Claude Mythos 5.1 is for trusted access programs.
- Mythos 5.1 safeguards support cybersecurity and life sciences.
- Fable 5.1 costs 25% less than Fable 5 for typical workloads.
- Fable 5.1 costs up to 45% less for highly agentic work.
- Enterprise Frontier Safeguards (EFS) provide complete data privacy.
- Fable 5.1 was released three months after Fable 5.
- Fable 5.1 sets a new standard on coding, knowledge work, and long-running problem-solving tasks.
- Fable 5.1 found the cause of a rare crash for Millennium.
- Project Glasswing launched in April.
- Project Glasswing gave early access to Claude Mythos Preview and Mythos 5.
- Claude Fable 5 is a broadly available model that blocks dual-use cyber work.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Primary sources
GitHub aws-samples/anthropic-on-aws GitHub harbor-framework/terminal-bench-science GitHub JohnHeibel/ClaudeAnimationBase GitHub anthropics/claude-code GitHub cmc_internal/api GitHub github/collectHow outlets covered it
A new benchmark called Livenerf has been launched to deterministically track the performance of Anthropic's Claude Opus 5.5 model over time, specifically to detect any potential degradation after its release. This initiative provides a day-0 baseline to statistically measure if large language models experience a 'nerf' or decline in quality post-launch, addressing community concerns about model consistency.
Anthropic released Claude Opus 5.5, positioning it as a lower-cost alternative to Fable 5.1 with comparable performance on many tasks. Benchmarks show Opus 5.5 outperforming Fable 5.1 in coding-related tasks, despite Anthropic's documentation still recommending Fable 5.1 for demanding reasoning. This analysis investigates the actual performance parity and cost-effectiveness of Opus 5.5 compared to Fable 5.1.
Anthropic released Claude Sonnet 5.5, incorporating cyber safeguards and model fallbacks previously reserved for its most capable models. This integration is due to Sonnet 5.5's increased proficiency in offensive security tasks, despite not advancing overall model capabilities.
Anthropic released Claude Sonnet 5.5, replacing Sonnet 5 as its medium-sized AI model, following the recent update to Opus 5.5. This new version offers improved performance for everyday tasks, faster output generation, and integrates cybersecurity safeguards previously reserved for more capable models.
Anthropic launched Sonnet 5.5, a new AI model that offers improved coding, task completion, and document creation at a lower cost than its predecessor. This release is Anthropic's second since its CEO advocated for a slower pace in AI development, emphasizing routine tasks over advanced capabilities.
Anthropic released Sonnet 5.5, an update to its mid-tier AI model, which is 30% faster and has a slower token burn rate than its predecessor. This model is positioned as an efficient assistant for tasks like coding and document creation, and it now includes cyber safeguards previously reserved for higher-tier models.
Anthropic released Claude Sonnet 5.5, an updated version of its AI model that provides performance close to Claude Opus 5.5 while being 30% faster and up to 30% cheaper than its predecessor. This update makes advanced AI capabilities more accessible and efficient for everyday tasks, bug fixing, and document creation.
Anthropic released Claude Opus 5.5, claiming it costs 40% less and generates output 30% faster than Opus 5. However, testing on complex reasoning tasks showed Opus 5.5 did not outperform its predecessor, despite the price reduction for API usage.
Anthropic has released Claude Opus 5.5, a new AI model that achieves performance levels comparable to Claude Fable 5.1 but at a 40% lower operational cost than Opus 5. The model features improvements in agentic coding, security, and communication, with zero data retention available.
Anthropic released Claude Opus 5.5, reducing pricing for input and output tokens. However, the update includes four breaking changes that can cause existing agents built for Opus 5 to fail, requiring code modifications.
Anthropic released Opus 5.5 and OpenAI released GPT-6 Sol and Luna, new AI models designed to offer improved efficiency and lower operational costs. These updates aim to make advanced AI more accessible and competitive against open-weight alternatives for enterprise users.
Anthropic released Claude Opus 5.5, offering performance comparable to Claude Fable 5.1 with a 40% cost reduction and 30% faster output. OpenAI launched GPT-6 Sol & Luna, which are 50% cheaper than GPT-5.6 models, making advanced AI more accessible and efficient for users.
Anthropic introduced Claude Opus 5.5, a new AI model designed to handle complete coding tasks from design to debugging and testing. This release is the first in the Claude 5.5 family, offering performance comparable to Claude Fable 5.1 on most work tasks at a 40% lower cost than Opus 5. The model aims to improve developer efficiency by completing more complex projects with fewer steps and tokens.
Anthropic released Opus 5.5 and OpenAI launched GPT-6 Sol and GPT-6 Luna, new AI models that offer performance improvements and reduced pricing. These updates aim to address enterprise needs, developer tasks, and general efficiency while making AI model usage more economical.
Anthropic's Claude Opus 5.5, the first in the Claude 5.5 model family, is now available on Amazon Bedrock and Claude Platform on AWS. This new version offers improved efficiency, lower per-token pricing, and enhanced clarity for agentic coding, knowledge work, and long-running tasks.
Anthropic has released Claude Opus 5.5, an upgraded version of its largest AI model, two months after the previous iteration. This new model demonstrates significant performance improvements, particularly in complex coding tasks and large-scale code migrations, offering increased efficiency and reduced costs compared to prior versions and competing models.
Anthropic launched Claude Opus 5.5, reducing its pricing by 20% and introducing a fast mode for Claude Code and the Claude Platform. The new model offers efficiency gains, including fewer tokens per task and faster output generation, which could impact model selection for agentic coding.
Anthropic released Opus 5.5, its most capable AI model, which offers improved performance in coding and knowledge work, surpassing the larger Fable model in some benchmarks. The new version also features reduced pricing for output tokens and faster operation, while incorporating similar safety safeguards as previous models.
Anthropic has launched Claude Opus 5.5, an updated AI model that matches the performance of Fable 5.1 while reducing operational costs by 40%. This release also includes a 20% reduction in token pricing and increased usage limits for subscribers, aiming to improve efficiency and accessibility for users.
Independent testing of Anthropic's Claude Fable 5.1 and Fable 5 on selected Terminal-Bench-Science tasks under budget-constrained conditions revealed less performance difference than Anthropic's official benchmark scores. This analysis suggests that the significant performance gains reported by Anthropic may not be fully realized by average users operating under typical resource limitations.
AWS has made Anthropic's Claude Fable 5.1 available on its platform, offering advanced AI capabilities for complex tasks with enhanced data governance. Additionally, AWS launched a public preview of Amazon Linux 2027 and new memory-optimized EC2 R9g/R9gd instances powered by Graviton5 processors. These updates provide AWS users with more powerful AI models, an updated cloud-native operating system, and improved performance for memory-intensive workloads.
An analysis compared Anthropic's new Claude Fable 5.1 with its predecessor, Fable 5, on practical tasks, finding no discernible performance difference despite benchmark improvements. This suggests that benchmark scores may not always reflect real-world utility for users.
Anthropic has launched new AI models, Claude Fable 5.1 and Mythos 5.1, addressing customer feedback on cost, data retention, and safeguards. Fable 5.1 offers improved performance and is up to 45% cheaper for agentic tasks, while Mythos 5.1 is available through Project Glasswing. These updates aim to make Anthropic's models more accessible and practical for enterprise use.
Anthropic has released Fable and Mythos 5.1, updated versions of its advanced AI models, featuring reduced token costs and fewer false-positive restrictions. The Fable model is now generally available, while Mythos 5.1 remains exclusive to partners in cybersecurity and life sciences research, and a new zero-data-retention service will roll out for enterprise clients.
Anthropic has launched Fable 5.1 and Mythos 5.1, updated versions of its flagship AI models, offering stronger performance and a 75% reduction in cache read pricing. Fable 5.1 is generally available, while Mythos 5.1 is restricted to a trusted access program for cybersecurity and life sciences applications. These updates aim to provide more capable and cost-effective AI solutions for users.
Anthropic has released Claude Fable 5.1 on Amazon Bedrock and Claude Platform on AWS, offering improved reasoning capabilities for coding, scientific research, and enterprise workflows. This version is designated a "Covered Model," subjecting it to additional data retention, safety review, and access policies, and introduces Enterprise Frontier Safeguards for customer data control.
Anthropic launched its new large language models, Claude Fable 5.1 and Mythos 5.1, which offer improved performance in agentic tasks and knowledge work. The release also includes a 75% cost reduction for cached context reads and a new security architecture, Enterprise Frontier Safeguards (EFS), allowing organizations to control monitoring data.
Anthropic has released Claude Fable 5.1 and Mythos 5.1, updated versions of its most capable AI models. The new models offer improved performance in coding, knowledge work, and problem-solving, along with reduced costs for typical and agentic workloads.
Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, new AI models designed for coding and knowledge work, offering performance improvements and cost reductions. Fable 5.1 is generally available, while Mythos 5.1 is for trusted access programs with specialized safeguards for cybersecurity and life sciences.
Anthropic is expanding access to its Mythos 5 AI model for cybersecurity applications through partner integrations and an updated Claude Security offering. The company also launched a $35 million Defender Advantage Fund to support open-source project security. This initiative aims to provide advanced AI defensive capabilities to cybersecurity teams while controlling direct model interaction to prevent misuse.
Claude Mythos 5, an AI model for cybersecurity, is now available in Claude Security and will integrate into partner tools. The company also launched a $35 million fund to support open-source software security and plans to expand its Cyber Verification Program. These actions aim to broaden access to advanced AI for defensive cybersecurity while maintaining safeguards against misuse.