# Helicone is maintenance-only: what to use instead (2026)

> Helicone went maintenance-only after its March 2026 Mintlify acquisition. What that means for your LLM cost tracking, and the strongest alternatives by use case.
> Published 2026-08-07 — https://cutmyaispend.com/blog/helicone-maintenance-mode-alternatives
Helicone spent three years as the default answer to "how do I see what my LLM calls cost?" — a one-line proxy swap that gave you per-request cost dashboards. In March 2026 it was acquired by Mintlify, and the project moved to **maintenance-only**: security fixes and new-model pricing continue, but feature development has stopped.

## What maintenance-only means for you

If you already run Helicone, nothing breaks. Existing deployments keep working, model price tables keep updating, and there is no forced migration. The risk is directional: observability tooling needs to track a fast-moving target (new providers, new token types like cached and batch tiers, new agent frameworks), and a frozen roadmap falls behind roughly one provider announcement at a time.

If you are choosing a cost-tracking layer **today**, starting on a maintenance-only product is hard to justify.

## Alternatives by situation

**You want cost tracking built into your gateway (most teams).** [LiteLLM](/tools/litellm) logs per-key, per-team, and per-model spend natively, and you probably want a gateway anyway for [model routing](/methods/model-routing) and [prompt caching](/methods/prompt-caching). Self-hosted, open source.

**You want managed, with caching included.** [Portkey](/tools/portkey) pairs cost analytics with the strongest built-in [semantic caching](/methods/semantic-caching), so the tool that measures your spend also cuts it.

**Your AI bill lives inside a bigger cloud bill.** [nOps](/tools/nops) folds Bedrock and Azure OpenAI spend into full-stack FinOps, which is the right frame when finance wants one dashboard for cloud plus AI.

**You mostly need attribution discipline, not another tool.** The uncomfortable truth from our [cost attribution guide](/methods/cost-attribution-finops): 73–79% of enterprises blew their 2026 AI budgets, and the common root cause was untagged spend, not missing dashboards. Metadata tagging through whatever gateway you already run covers most of what Helicone did.

## The bigger picture

Cost *visibility* is step one; it does not cut anything by itself. Once you can see per-feature spend, the ranked playbook on [our homepage](/) is the order of operations — and the deepest cut is usually [fixing redundant context at the source](/methods/fix-the-context-layer) rather than discounting it after the fact.
## Sources

- nOps: 10 Best LLM Cost Optimization Tools in 2026: https://www.nops.io/blog/llm-cost-optimization-tools/
- Maxim: 5 Tools for Reducing LLM API Costs in Production (2026): https://www.getmaxim.ai/articles/5-tools-for-reducing-llm-api-costs-in-production-2026/