}
        logger.info("[LLM] {} initialized. model={}, timeout={}ms, maxConcurrent={}", getName(), getModel(), getTimeout(),
                getMaxConcurrentRequests());

        concurrencyLimiter = new Semaphore(getMaxConcurrentRequests());

        startAvailabilityCheck();
    }

    /**
     * Cleans up resources.
     */
    public void destroy() {