whereismyai
ScaledThought
ScaledThought is an API service providing open-weight models through an OpenAI-compatible interface set up via command-line agents.
ScaledThought provides access to open-weight models through a single OpenAI-compatible application programming interface. The service is designed to be set up automatically by command-line agents such as Claude Code, Codex, or Cursor using a single prompt or command. Users pay per token using prepaid credits rather than monthly subscriptions, with top-ups ranging from twenty to two thousand dollars approved directly by humans via secure links. The platform hosts models in the United States region with Canada coming soon. It also supports LoRA fine-tuning where jobs are billed per training token and refunded in full if cancelled or failed. Unused credit balances do not expire and requests halt with a four hundred two status code when funds deplete.
ScaledThought at a glance
- Best for: Software engineer
- Runs on: API, Web, CLI
- Pricing: No subscription; pay-per-token from prepaid credits starting at $20 top-ups up to $2,000 per 24 hours.
- Released: October 2026
- Category: AI coding tool
- Website: scaledthought.com
Key features
- OpenAI-compatible API
- Agent-driven setup
- Per-token pricing
- Prepaid credit system
- LoRA fine-tuning
- Automatic prompt caching
Use it for
- Running open-weight models via an OpenAI-compatible API
- Setting up model access and keys through an AI agent
- Fine-tuning models using LoRA with training token billing
- Managing API usage and budgets with prepaid credit top-ups
Alternatives to ScaledThought
- iwant — Spins up cloud GPUs and open models instantly via a single command.
- OpenCode — Works with 75+ model providers, local models, and existing subscriptions
- Tokscale — Track and compare token usage across multiple AI coding agents.
- OpenRouter — Access over 500 models and 80 providers through one unified API.
- pstack — Improves AI coding agent outcomes with rigorous workflows.
- Replicate — Run and fine-tune thousands of open-source and custom AI models via a single cloud API.
- OpenSandbox — Provides isolated environments with multi-language SDKs and lifecycle management.
- oMLX — It uses paged SSD KV caching to reduce agent TTFT to under 5 seconds.
- Strata — Runs large AI models locally on consumer gaming PCs without server reliance.
- Rigmark — Measures local AI performance with reproducible, agent-like workloads.
- Freebuff — Access multiple coding AI models for free without subscriptions.
- OmniRoute — Routes requests across 352 providers with automatic fallback and cost-based selection.
All ScaledThought alternatives