about aiworkbench.dev

AIWorkbench.dev is a free, open-source developer toolkit for testing, comparing, and debugging large language models from Anthropic, OpenAI, Google Gemini, AWS Bedrock, DeepSeek, and Meta — all in one place.

our mission

We believe every developer deserves fast, transparent access to LLM capabilities without vendor lock-in, opaque pricing, or sending data through third-party servers. AIWorkbench runs entirely in your browser — your API keys never touch our infrastructure.

what we build

• ✓ API Workbench — full parameter control for any model
• ✓ Multi-Model Compare — one prompt, six providers at once
• ✓ Streaming Debugger — live TTFT, tokens/sec, stall detection
• ✓ Token Counter & Cost Calculator — no API key needed
• ✓ Caching Calculator — see exact prompt caching savings
• ✓ Prompt Library — save, version, and diff prompts

privacy & security

Your API keys are stored in sessionStorage only — never sent to our servers, never persisted to disk, never shared. All API calls go directly from your browser to the provider endpoint. We cannot see your prompts, your outputs, or your keys.

the team

AIWorkbench is built by a small team of independent developers passionate about developer experience and AI transparency. We are not affiliated with any LLM provider.