Run Terminal-Bench and other Harbor evals on Vercel Sandbox8m ago Nvidia's Huang diverges with CEOs of Anthropic, OpenAI on AI safety at Dreamforce26m ago AI text watermarking can make models more vulnerable to adversarial prompts35m ago AI needs to have 'reasonable guidelines,' Palantir's Karp tells CNBC36m ago Kris Beevers, CEO and Co-Founder, Netbox Labs – Interview Series44m ago Delta Unveils Energy-to-Compute Infrastructure for NVIDIA DSX AI Factories46m ago Fearing No Repercussions, OpenAI Admits That Its Rogue AI Agents Performed a Bunch of Other Terrifying Actions54m ago Pinterest launches Visual Search Ads at its advertiser summit57m ago Existential Angst Dominates Debate Over Tech From Anthropic, OpenAI57m ago Magentic raises $18M to put AI agents to work in manufacturers’ supply chains57m ago AI Pioneer Andrew Ng Calls Extinction Fears ‘Science Fiction’59m ago Reduce time-to-hire for quality candidates with AI-powered Amazon Connect Talent1h ago GitHub and Anthropic used their own agents for major Rust rewrites — with very different playbooks1h ago Nvidia’s Huang Expects to Sell Twice as Many Chips Next Year1h ago

Run Terminal-Bench and other Harbor evals on Vercel Sandbox8m ago Nvidia's Huang diverges with CEOs of Anthropic, OpenAI on AI safety at Dreamforce26m ago AI text watermarking can make models more vulnerable to adversarial prompts35m ago AI needs to have 'reasonable guidelines,' Palantir's Karp tells CNBC36m ago Kris Beevers, CEO and Co-Founder, Netbox Labs – Interview Series44m ago Delta Unveils Energy-to-Compute Infrastructure for NVIDIA DSX AI Factories46m ago Fearing No Repercussions, OpenAI Admits That Its Rogue AI Agents Performed a Bunch of Other Terrifying Actions54m ago Pinterest launches Visual Search Ads at its advertiser summit57m ago Existential Angst Dominates Debate Over Tech From Anthropic, OpenAI57m ago Magentic raises $18M to put AI agents to work in manufacturers’ supply chains57m ago AI Pioneer Andrew Ng Calls Extinction Fears ‘Science Fiction’59m ago Reduce time-to-hire for quality candidates with AI-powered Amazon Connect Talent1h ago GitHub and Anthropic used their own agents for major Rust rewrites — with very different playbooks1h ago Nvidia’s Huang Expects to Sell Twice as Many Chips Next Year1h ago

Dev1 min reading time

Run Terminal-Bench and other Harbor evals on Vercel Sandbox

VercelPublished 8 minutes ago

Harbor benchmarks like Terminal-Bench can now run on Vercel Sandbox using isolated Firecracker microVMs for high parallelism. The setup supports secure network policies and credential injection, enabling benchmarking across multiple AI models via AI Gateway.

#ai-gateway #benchmarking #firecracker #harbor #terminal-bench #vercel-sandbox

More in Dev

Dev6 min read

GitHub and Anthropic used their own agents for major Rust rewrites — with very different playbooks

The New Stack (AI)1h ago

Dev12 min read

A serverless, data-driven Git metrics dashboard using Amazon Quick Sight

AWS Blog3h ago

Dev1 min read

The skills CLI now supports Notion hosted skills

Vercel1h ago