RunInfra
Open models, built for agents
RunInfra offers hosted open-source large language model APIs built for AI agents, addressing the need for fast inference with lower-cost cached context. Session-aware routing keeps repeat calls close to cached context by preferring the replica holding a session's context, identified automatically from the conversation or workspace or via a session ID. Prompts and completions are not stored, and the service is SOC 2 Type II attested.
The platform provides a single API key compatible with OpenAI and Anthropic SDKs, with OpenAI-compatible chat completions and Anthropic-compatible Messages endpoints. The model library includes Nemotron 3.5 Lightning 30B, Qwen3.8 27B, Ornith 1.5 35B, GLM 5.3 Flash, and DeepSeek V4.1 Flash, with capabilities such as tool calling, JSON mode, streaming, and image input. Cache hits are reported in usage and billed at the cached rate. Workspaces support roles and row-level security, with audit logs and RBAC available on Business accounts and Enterprise plans. Setup with coding agents such as Claude Code, Codex, OpenCode, Cline, Droid, Grok, and Build is supported via a CLI.
RunInfra is aimed at developers and teams building AI agents and applications. Pricing is usage-based: pay-as-you-go credit top-ups with per-token billing and no per-seat fees, coding plans with usage that resets periodically, and an Enterprise edition with custom volume, dedicated infrastructure, custom SLAs, and SOC 2 Type II attestation. New accounts receive a small credit to test the Model APIs.
12 alternatives to RunInfra
Ranked by how well each tool replaces RunInfra: shared features, audience, price and popularity.
- 68 out of 100 matchUsage-based
- 66 out of 100 matchUsage-based
Cloud platform for running open-source machine learning models via API
Covers 2 of 15 key features and is open source.
Free planOpen source64 out of 100 matchUsage-based- 64 out of 100 matchUsage-based
- 63 out of 100 matchContact sales
Serve and scale open-source and custom AI models on the fastest, most reliable inference
Covers 3 of 15 key features.
Free plan63 out of 100 matchFreeThe unified interface for every model
Covers 3 of 15 key features and is open source.
Free planOpen source63 out of 100 matchFree- 63 out of 100 matchUsage-based
- 62 out of 100 match$9/mo
- 62 out of 100 matchUsage-based
Inference built for coding agents, with an OpenAI-compatible API and a toolkit of models
Covers 2 of 15 key features.
62 out of 100 match—