Futures
Access hundreds of perpetual contracts
TradFi
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
Launchpad
Be early to the next big token project
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
OpenClaw Introduces 'Dreaming' Memory System, Similarities with Leaked Claude Code Source
According to monitoring by 1M AI News, the open-source AI assistant platform OpenClaw has released version 2026.4.5, which includes approximately 40 new features and hundreds of fixes, completed by over 60 contributors. This is the first major version update after Anthropic banned OpenClaw’s subscription access. The most notable new feature is the experimental ‘dreaming’ memory system, which integrates backend memory into three collaborative stages: light sleep filters short-term memory, deep sleep is responsible for persistent writing, and the REM stage discovers associations and refines lasting insights. Each of the three stages has independent scheduling and can run automatically without user intervention, accompanied by a Dream Diary interface and debugging tools. The leaked Claude Code source from March 31 coincidentally includes a mechanism called autoDream, which also automatically integrates, deduplicates, and eliminates memory conflicts during idle times, employing a similar layered architecture: the index is always loaded, themes are read on demand, and history is searched without loading. Another similarity is in prompt cache optimization. This version of OpenClaw improves cache hit rates through deterministic arrangement of MCP tool definitions, normalization of system prompt fingerprints, and stripping of inline tool lists, while the leaked Claude Code source shows that it uses stable/dynamic prompt boundaries to maximize cache reuse, indicating nearly identical optimization directions. Boris Cherny, an engineer from Claude Code, is also one of the contributors to OpenClaw’s cache improvements. However, development records for OpenClaw’s related PRs indicate that the development cycle for these features began well before the leak, making it unrealistic to complete such a large-scale feature port in less than a week. A more likely scenario is that both projects arrived at similar solutions while addressing the same issues, and the leak provided the first opportunity for external comparison of their internal designs. Other major updates include: 1. Built-in video generation tools supporting xAI, Alibaba Cloud Wan, and Runway; music generation tools supporting Google Lyria and MiniMax. 2. New model providers including Tongyi Qianwen, Fireworks AI, and StepFun, with Amazon Bedrock adding Mantle auto-discovery. 3. The control panel now supports 12 languages including Simplified Chinese, Traditional Chinese, Japanese, and Korean. 4. Multiple security enhancements for Claude CLI integration, clearing inherited environment variables and configuration overrides, enforcing restricted permission modes to prevent third parties from hijacking the Claude authentication context through OpenClaw.