RunInfra
RunInfra is a RightNow product offering hosted APIs for open models, built for AI agents. It uses session-aware routing to keep repeat calls on warm replicas, reporting high cache hit rates and lower cached-input pricing, with one API key compatible with the OpenAI and Anthropic SDKs.
Key facts
What is RunInfra?
Fast, low-cost inference for agents by keeping sessions warm and billing repeated context at cached-input rates.
Who is RunInfra best for?
Developers and teams building AI agents; users of Claude Code, Codex, OpenCode and 20+ other coding agents
What does RunInfra cost?
Verified starting price is USD 10 / month. View source
What are its main limitations?
- Cache hits are explicitly best effort, so effective rates are not guaranteed
- Pay-as-you-go requires a minimum $5 top-up
- Enterprise features (SOC 2 Type II, SLA, dedicated support) are not included in self-serve
- Only 8 models available in the library
Limitations are based on product materials or editorial synthesis; confirm on the official site.
When was this information verified?
October 20, 2026
Sources
- Official homepageofficial homepage · Oct 20, 2026
- Official pricingofficial pricing · Oct 20, 2026
Key features
- ✓One API key for the OpenAI and Anthropic SDKs, with OpenAI-compatible chat completions and Anthropic-compatible Messages
- ✓Session-aware routing that sends repeat calls to the replica holding their context, so sessions stay warm
- ✓Reported cache hit rates up to 99.3% (last 24 hours) with cache hits shown in usage and billed at the cached rate
- ✓Cached input pricing from $0.01 per 1M cached input tokens on supported open models
- ✓Model library of 8 open models with per-model pricing, cache hit rate and throughput data
- ✓Coding plans for Claude Code, Codex, OpenCode and 20+ agents, with usage that resets every 5 hours and weekly
- ✓Pay-as-you-go prepaid balance with no seats and no per-user fees; top-ups valid for one year and optional auto-recharge
- ✓Audit logs and role-based access control (RBAC) available free with a Business account
Use cases
- →Running agent loops that resend the same context every turn at a reduced effective input price
- →Powering coding agents such as Claude Code, Codex and OpenCode with a subscription plan
- →Comparing model cost, throughput and cache hit rates before choosing an open model
- →Migrating existing OpenAI or Anthropic SDK code to open models via a compatible endpoint
- →Enterprise deployments requiring SOC 2 Type II, custom volume and SLAs up to 99.99%
Pricing
Pay as you go
$5 minimum top-up
- · Per-token pricing at published model prices
- · Credit top-ups from $5, no subscription
- · Credits last one year, optional auto-recharge
- · OpenAI- and Anthropic-compatible APIs
- · No per-seat fees
- · Get $1 free to test the Model APIs
Coding plan Starter
$10/mo
- · For Claude Code, Codex, OpenCode and 20+ agents
- · Usage included, resets every 5 hours and weekly
- · At a limit: credits or free Standby
Coding plan Pro
$29/mo
- · Usage included, resets every 5 hours and weekly
- · At a limit: credits or free Standby
Coding plan Team
$99/mo
- · Usage included, resets every 5 hours and weekly
- · At a limit: credits or free Standby
Enterprise
Custom
- · Dedicated infrastructure and reserved capacity
- · Audit logs and role-based access control (RBAC)
- · Custom volume and contract terms
- · Custom SLAs up to 99.99%
- · SOC 2 Type II attestation
- · Dedicated CSM and private Slack
Pay-as-you-go credit top-ups from $5 with per-token model pricing; coding plans from $10/mo (Starter $10, Pro $29, Team $99); custom Enterprise pricing.
Pros
- +Session-aware routing produces very high measured cache hit rates (up to 99.3%)
- +OpenAI- and Anthropic-compatible APIs reduce integration work
- +No subscription or per-seat fees on pay-as-you-go; credits valid for one year
- +Transparent per-model pricing, throughput and cache hit data in the model library
- +Backed by Y Combinator
Cons
- −Cache hits are explicitly best effort, so effective rates are not guaranteed
- −Pay-as-you-go requires a minimum $5 top-up
- −Enterprise features (SOC 2 Type II, SLA, dedicated support) are not included in self-serve
- −Only 8 models available in the library
Positioning
- Core value: Fast, low-cost inference for agents by keeping sessions warm and billing repeated context at cached-input rates.
- Ideal for: Developers and teams building AI agents; users of Claude Code, Codex, OpenCode and 20+ other coding agents
- Product type: Web app and API (OpenAI-compatible chat completions and Anthropic-compatible Messages)
Frequently asked questions
What is RunInfra?
RunInfra is a hosted inference platform from RightNow that serves open models through APIs built for agents, using session-aware routing and cached-input billing.
How much does RunInfra cost?
Pay as you go uses credit top-ups from $5 with per-token pricing (e.g. GLM 5.3 Flash at $0.11 per 1M input tokens, $0.03 cached input, $0.45 output). Coding plans are Starter $10, Pro $29 and Team $99 per month. Enterprise is custom priced.
Is there a free plan?
The site lists no free plan tier, but offers $1 free credit to test the Model APIs, and coding plans fall back to credits or a free Standby mode when a usage limit is reached.
Which SDKs are compatible with RunInfra?
One API key works with the OpenAI and Anthropic SDKs; the platform offers OpenAI-compatible chat completions and Anthropic-compatible Messages.
Do credits expire?
Top-ups stay valid for one year, and optional auto-recharge adds credits automatically when your balance runs low.
You may also like
Curated suggestions based on similar categories.
Asimov 1
Asimov 1 is a 1.2 m, 35 kg humanoid robot built for whole-body control and locomotion research. It is sold as developer-friendly, open hardware that supports sim-to-real policy transfer, raw data collection, and live teleoperation via joystick, code, or an agent. Orders are started with a $499 deposit, with full pricing handled through sales or invoice request.
Exa 3.0
Exa 3.0 is the latest release of Exa's web search API for AI applications, featuring a larger embedding model, generally available Dynamic Highlights, and a substantially expanded index covering ecommerce, local businesses, news, sports, weather, flights, events, transit, and podcasts. The release introduces Exa Lite, a lower-cost search type priced at $4 per 1,000 requests, alongside Exa Auto at $7 per 1,000 requests.
Harness²
Harness² (harness2) is a Google Research plug-and-play framework for recursive agent harnessing in an open world. It keeps model weights fixed while optimizing the harness around a model across two levels of recursion — task-level and harnessing-level — through eight editable harness components. Released under the Apache 2.0 license.
Beam.win
Beam is an AI platform built for human services teams, offering AI-assisted case notes (Notes), live two-way translation (Interpret), and voice agents that handle inbound calls (Talk). It is used by frontline government, non-profit and welfare organisations to reduce administrative workload.
Step 5 Preview
Step 5 Preview is StepFun's flagship model for agentic work, built on a sparse Mixture-of-Experts architecture with 600B total parameters (27B active per token), a 1M-token context window, and vision input. It targets software engineering, long-horizon agentic tasks, professional knowledge work, and finance, and is available through StepFun's products and API, with open weights planned for October 15.
Nemotron 3.5 Super VL
Nemotron 3.5 Super VL is a member of NVIDIA's Nemotron 3 family of open models with open weights, training data, and recipes. It combines visual understanding, reasoning, and tool use in a single model and is described as delivering highest in-class efficiency and accuracy for specialized domains. It can be deployed from local infrastructure to the cloud and is accessible via OpenRouter, NVIDIA NIM API, and Hugging Face.