Back to Ecosystem Pulse
EvaluationExperimental
LMMOCK
by lewismosciski
Local deterministic mock for OpenAI and Anthropic APIs
Python
Updated Oct 1, 2026
Share:
What It Does
Provides a lightweight local mock LLM server that returns deterministic OpenAI- and Anthropic-compatible responses without making API calls. It runs a FastAPI-based HTTP mock that mirrors OpenAI/Anthropic endpoints so existing clients and agent code work unchanged. The server removes tokens and network dependencies, enabling repeatable, high-speed tests and offline development. local mock LLM server
The Value Proposition
As multi-agent systems become more interconnected, deterministic, repeatable LLM responses are essential for reliable evaluation and debugging. LMMOCK lets teams run end-to-end agent-to-agent scenarios end-to-end agent-to-agent scenarios and regressions without flakiness from remote APIs or cost/noise from tokens. That predictability is a simple but crucial building block for continuous agent evaluation, failure-mode testing, and tracking agent behavior over time. regressions
Ideal For
Developers and QA teams who need repeatable, offline LLM responses to test agents, pipelines, and integration logic without changing code. repeatable, offline LLM responses
How It's Used
- Run deterministic integration tests for multi-agent workflows that call OpenAI/Anthropic APIs
- Reproduce and debug agent failure modes without incurring API costs or rate limits
- Perform continuous agent evaluation in CI with stable, predictable responses
Works With
openaianthropicfastapipython
Topics
agent-testingai-testinganthropicfastapillmmock-serveropenaipython
Similar Tools
openai-mockmock-llm
Keywords
mock-llmpre-production agent testingagent reliabilityopenaianthropic