Coinbase Blog
Featured
Most recent
Coinbase and Thunes connect stablecoin infrastructure with global fiat payouts
TL;DR: We’re collaborating with Thunes to connect our stablecoin infrastructure with its global fiat payout network. Coinbase will power Thunes’ stablecoin infrastructure, while Thunes will expand Coinbase’s payout reach to local bank accounts and mobile wallets. Together, we’ll support cross-border payments across enterprise supplier payments, gig worker payouts, and international remittances.
An Update to Our Asset Listings Announcements
TL;DR: @CoinbaseAssets is now the dedicated source for all official Coinbase asset announcements on X. @CoinbaseMarkets will remain focused on traders, including market insights, product updates and trading education, while amplifying asset moments that are relevant to its audience.
Discover more
Read and watch the latest ecosystem news, insights, and resources
Singapore Sees 43% Drop in Crypto Scam Losses and Partnerships Like Ours Are Helping Drive the Change
TL;DR: Singapore’s crypto scam losses fell 43% year over year, from S$114.9 million to S$65.5 million, as a result of stronger enforcement, public education, and partnerships between the Anti-Scam Centre and industry leaders like Coinbase. Through four joint operations, Coinbase’s Global Intelligence team helped identify 720+ victims and prevent more than S$18.9 million in potential losses, while a new anti-scam awareness campaign reached Singapore retail customers.
Part 3/3 Owning Intelligence Series: How Coinbase Built, Evaluated, and Post-Trained a Fraud Agent
TL;DR: Using SkyRL on Anyscale, we post-trained Qwen3.5-9B for Onramp fraud detection and served it with Ray Serve LLM and vLLM. On our benchmark, it exceeded Opus 4.5 across all four detection metrics, including 9.6% higher F1. In online inference, we reduced end to end latency by 55%.
Part 2/3 Owning Intelligence Series: How Coinbase Built, Evaluated, and Post-Trained a Fraud Agent
TL;DR: Newer LLMs are not necessarily better at domain-specific tasks. On our fraud benchmark, newer Opus, Sonnet, and GPT versions all had lower recall, F1, and dollar-weighted recall—even when precision improved. The lesson: choose models using high-quality, domain-specific benchmarks, with reproducible evaluation and ideally deterministic scoring against verified outcomes. These findings also motivated our next step: post-training open-weight models for greater control over quality, cost, and deployment.
