Cloud starter
20 introductory Boosts free
See stated billing terms
- Allowance and restrictions
- Twenty introductory Boost episodes; no card required; recurring refill and expiry unconfirmed.
Apple Silicon local LLM toolkit · Official-source review 2026-10-09 UTC
OptiQ is an MLX-based toolkit for quantizing, fine-tuning and serving language models on Apple Silicon, with a local workbench and an optional cloud escalation.
Understand the total cost
Cloud starter
20 introductory Boosts free
See stated billing terms
Cloud Boost pack
$20 once
See stated billing terms
Local toolkit
PyPI distribution; verify licence
See stated billing terms
No retained pricing changes yet. A current price alone does not establish a historical trend.
Product identity and relevant official-source evidence reviewed individually.
Read original source ↗Cloud Boost costs USD20 for two hundred episodes, not a monthly subscription; twenty starter Boosts are introductory. One episode may contain several tool-loop calls; no-output episodes advertise automatic refund. Auto top-up is off by default but can create paid purchases when enabled. Credit expiry and software/model licensing terms unconfirmed. Apple Silicon M1–M5, Python3.11+ and sixteen-GB-plus RAM examples; actual memory depends on model. Cloud Boost transmits episode context despite local routine inference. Quantization benchmarks, sandbox security and coding quality were not tested; nothing installed or executed.
Read original source ↗One Boost per episode
Read original source ↗Purchased Boosts are a one-off pack, not a subscription.
Read original source ↗Off unless you turn it on.
Read original source ↗Facts apply to the named version and check date. Send a sourced correction if something changed.
OptiQ offers a way to start free, but not a standard ongoing plan. Twenty introductory Cloud Boosts without a card. Local toolkit distributed on PyPI; local hardware/model licence costs and toolkit licence terms require checking, rather than assuming unrestricted free use.
OptiQ does not publish a monthly price we could confirm.
API or developer access is mentioned. Local inference server advertises OpenAI- and Anthropic-compatible protocols; CLI, Lab and coding agent share it. Optional cloud Boost sends a task episode to a frontier model.