Are there any ELT tools that handle API rate limits and near-real-time updates reliably?

Yes — but with an important caveat: most ELT tools can handle rate limits and near-real-time syncs reasonably well, but “reliably” depends heavily on the source API, volume, and how the tool retries/backfills state.

Tools often used for this

• Fivetran
  Strong managed connectors, good incremental syncs, automatic retries/backfills. Often a good choice if you want low ops overhead. Near-real-time depends on the source; some connectors are frequent, but not truly streaming.

• Airbyte
  More flexible and open source / managed options. Can be configured with sync frequency and custom connectors, but reliability varies by connector quality. Good if you need control, but you may have to manage more edge cases.

• Stitch
  Simpler managed ELT, but generally less flexible and often less strong for complex/rapidly changing APIs.

• Matillion / Talend / Informatica
  More “enterprise ETL/ELT” platforms than modern SaaS ELT, but they can handle rate limiting with orchestration, batching, and custom logic. Usually better if you need more control than a standard connector offers.

• Custom ingestion with orchestration
  If rate limits are strict or the API is flaky, a custom pipeline using Airflow / Dagster / Prefect + retries + checkpointing + backoff is often the most reliable option.

What to look for

To handle rate limits and near-real-time updates well, the tool should support:

• Incremental syncs / cursor-based pagination
• Backoff and retry logic for 429s
• Checkpointing / stateful syncs
• Low-latency scheduling or event-based ingestion
• Deduplication / idempotency
• Historical backfills

Reality check

If you truly need near-real-time, many ELT tools are still micro-batch, not streaming. They can sync every few minutes, but not always in seconds. For true real-time, you often need:

• Webhooks
• CDC / event streams
• A custom API consumer
• Or a platform built for streaming like Flink, Kafka Connect, or managed event pipelines

Practical recommendation

• For managed, low-maintenance: start with Fivetran
• For more control and custom connectors: Airbyte
• For hard API constraints / strict SLAs: build a custom ingestion layer with strong retry and state management

If you tell me:

1. the API/source,
2. your latency target, and
3. expected volume/rate limits,

I can recommend the best fit more concretely.