Skip to content
ATAI Today Brief
HomeNewsConceptsGuidesToolbox
AboutSubscribeUA
Subscribe

AI Today Brief

The daily AI-engineering brief. Built in public. EN · UA.

XTelegramLinkedInYouTubeRSS

Follow AI Today Brief on LinkedIn for daily AI-engineering updates and the weekly “5 shifts that changed how developers work” PDF.

Explore

NewsDigestsConceptsGuides

Company

SubscribeAdvertiseAbout

Legal

Editorial policyAI disclosurePrivacyTerms

© 2026 AI Today Brief. All rights reserved.

  1. Home/
  2. News/
  3. Agents & MCP/
  4. Sakana AI Releases Fugu-Cyber Model Endpoint for Automated Security Workflows
Agents & MCP

Sakana AI Releases Fugu-Cyber Model Endpoint for Automated Security Workflows

July 26, 2026· 4 min read
OKCurated by Oleksandr Kuzmenko, AI Product Engineer·Updated July 26, 2026·Sources cited on every story
AI-assisted · editor-reviewed·How we use AI
Sakana AI Releases Fugu-Cyber Model Endpoint for Automated Security Workflows

Sakana AI launched Fugu-Cyber (fugu-cyber-v1.0), a specialized cybersecurity endpoint for its Fugu agent orchestrator. The model scored 86.9% on the CyberGym vulnerability verification benchmark and 72.1% on Microsoft's CTI-REALM detection benchmark.

Impact: Medium

Why it matters

Engineers can automate vulnerability validation and Sigma rule creation using multi-agent verification workflows.

TL;DR

  • 01Multi-agent verification workflows reduce false positive security alerts in automated vulnerability checks.
  • 02Large codebase analysis over 272K tokens incurs a 100% price surge under Sakana's tiered pricing model.

Key facts

CyberGym Benchmark86.9% (self-reported)
CTI-REALM Score72.1% (self-reported)
Model ID
fugu-cyber-v1.0
CyberGym Benchmark
86.9% (self-reported)
CTI-REALM Score
72.1% (self-reported)
Input Pricing
$6.00 / 1M tokens
Output Pricing
$36.00 / 1M tokens
Context Penalty Threshold
2x price tier above 272K tokens

Agentic Security Benchmark Performance

Sakana AI released fugu-cyber-v1.0, a cybersecurity-focused endpoint for the Fugu orchestrator. In testing, the model reached an 86.9% success rate on UC Berkeley's CyberGym benchmark (comprising 1,507 vulnerabilities across 188 OSS-Fuzz projects) and 72.1% on Microsoft's CTI-REALM threat detection benchmark.

TRINITY Architecture and Verification Routing

Fugu-Cyber uses the TRINITY framework to separate responsibilities into Thinker, Worker, and Verifier agent roles. Sub-agents execute vulnerability proof-of-concept checks against codebases to eliminate false positives.

Pricing Tiers and Context Penalties

Base pricing is $6.00 per million input tokens, $36.00 per million output tokens, and $0.60 per million cached input tokens. Once a run crosses the 272K token context limit, all pricing tiers double. Access requires manual approval and is restricted in the EU/EEA.

Try it in 2 minutes

curl -X POST https://api.sakana.ai/v1/chat/completions \
  -H "Authorization: Bearer $SAKANA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "fugu-cyber-v1.0", "messages": [{"role": "user", "content": "Verify vulnerability in unpatched codebase"}]}'

bash

✓ When to use

  • Automating vulnerability PoC generation across open-source codebases.
  • Converting threat intelligence reports into actionable Sigma detection rules.

✕ When NOT to use

  • Offensive penetration testing prohibited by Sakana Acceptable Usage Policy.
  • Deploying production workloads within the EU or EEA.

What to do today

  • →Apply for API access through Sakana AI's application form if building defensive security tooling.
  • →Monitor context length in automated sweeps to avoid crossing the 272K token price doubling threshold.
#Fugu-Cyber#Claude Mythos#GPT-5.5-Cyber#TRINITY

Sources

  • Sakana AI Releases Fugu-Cyber Orchestration Model
ShareShare on XShare on LinkedIn
← Previous storyOpenSlides Offers Open-Source Offline Code Presentation App Built with Svelte 5 and RustNext story →Publicly Shared Claude Conversations Expose Embedded API Keys and Secrets to Search Engines

Related stories

  • Agents & MCPVending-Bench Exposes Autonomous Agent Collusion and Price Fixing in Business Simulations
  • Agents & MCPStructuring Claude Subagent Instructions for Minimal Context Overhead
  • Agents & MCPOfficial Model Context Protocol C# SDK v2.0 Releases with Stateless HTTP Transport
  • Agents & MCPNVIDIA Open-Sources NOOA: Python Object-Oriented Framework for AI Agents

Email digest

Get the morning AI brief

One email a day — the stories that matter for engineers, founders and tech leads. Human-edited, with links to primary sources.

  • ✓120+ sources scanned daily
  • ✓Edited by a human
  • ✓1 email per day
  • ✓EN + UA

By subscribing you agree to the privacy policy.