FlashLabs Launches Claude Fable 5 API via OrcaRouter ― Mythos-level Performance with Up to 10% Bonus Credits for Subscribers

On Tuesday, June 10, 2026, FlashLabs, based in Chiyoda-ku, Tokyo, led by CEO Yoichi Hosoi, announced the launch of Anthropic's latest AI model, 'Claude Fable 5,' through their partner next-gen AI inference gateway, 'OrcaRouter.' Concurrently, the company is running a campaign offering up to 10% bonus credits for monthly subscription users.

Background and Objectives As enterprise AI adoption accelerates, companies are pressured to choose from over 200 LLM models. Using a single high-performance model for all processes leads to unnecessarily high costs for routine tasks, causing a spike in AI expenditures. Furthermore, manual model selection imposes significant maintenance burdens on development teams as new model releases render existing rules obsolete.

Overview of Claude Fable 5 × OrcaRouter - Launch Date: June 10, 2026 (JST) - Provisioning Method: API access via OrcaRouter

### Claude Fable 5 Key Specifications: - Pricing: $10 per 1M input tokens, $50 per 1M output tokens (over 50% discount compared to Mythos Preview) - Performance: Mythos-level peak performance; top-tier in software engineering, knowledge work, vision, and scientific research - Safety: General-purpose version equipped with robust protection against misuse - Environment: Claude API (model name: claude-fable-5)

### OrcaRouter Compatibility: Unifies API calls to over 200 LLMs from 15+ companies—including OpenAI, Anthropic, Google, xAI, Meta, Mistral, DeepSeek, Alibaba, Moonshot, and ByteDance—into a single endpoint, a single API key, and a single invoice.

Value for Enterprises 1. Mythos-level Performance at Lower Cost: OrcaRouter's adaptive inference gateway automatically routes prompts based on difficulty, utilizing Claude Fable 5 for complex tasks while routing routine tasks to lightweight models. 2. Automated Model Adaptation: The 'Smart Warm-up' feature optimizes the balance between exploration and exploitation, maintaining quality while adapting to real-world traffic. 3. 5-Minute Integration: Built as an OpenAI-compatible API, it works with existing OpenAI SDK code without modification by simply updating the Base URL and API key.

OrcaRouter Technical Features OrcaRouter frames LLM routing as a 'contextual bandit problem,' employing an Embedding-enhanced LinUCB algorithm. This approach enables adaptive model selection by combining offline learning with online feedback.