Helicone

Helicone is an open-source LLM gateway and observability layer. Extra request logs start around $0.28 / 100k. Helicone/helicone showed about 4,500 stars. The old "8,000+ stars / 2,000+ companies / free + paid unspecified" line is stale.

Compare Langfuse ($0 / $29 / $199 seats) if you wanted evals plus self-host, and LangSmith if you wanted LangChain Cloud.

Key Features

• Proxy or SDK: Point base_url at Helicone, or use their SDKs.
• Logs, cost, cache, limits: Gateway features sit next to traces.
• Self-host: OSS remains an option. Cloud is the priced grid above.
• Not 8k stars: Quote the dated GitHub count.

Use Cases

• Teams that want a gateway, not only a trace UI.
• People caching repeated prompts to cut provider spend.
• Builders comparing $79 Pro vs Langfuse Core $29.

Limitation: included log counts differ by plan and were not all copied. Confirm the meter. Fine-tune marketing should be confirmed in current docs.

Pricing

helicone.ai/pricing on 2026-08-17:

Getting Started

1. Read helicone.ai or the GitHub README.
2. Change one OpenAI/Anthropic client to the Helicone base URL.
3. Stay on Free until you know log volume.
4. Self-host only if you will operate the stack.

Frequently Asked Questions

Is cloud still "free + paid"?

No. Pro is $79 and Team is $799.

8k stars still right?

No. About 4,500 on 2026-08-17.

Helicone or Langfuse?

Helicone if you want the proxy/gateway. Langfuse if you want evals and datasets first.

Alternatives

• Langfuse: Traces and evals, $29 / $199 seats.
• LangSmith: LangChain Cloud, $39 Plus.
• OpenAI: Provider-native usage dashboards if you stay on one lab.

Tips

1. Quote $79 / $799 and $0.28 / 100k together.
2. Date the star count.
3. Confirm cache behavior. It changes the bill more than the seat.

Conclusion

Helicone is a priced OSS gateway with public $0 / $79 / $799 cloud seats, not an 8k-star slogan. Start at helicone.ai/pricing, then compare Langfuse.