Arkor
Arkor is a platform that hosts the entire loop for open-weight models: an OpenAI-compatible endpoint, request logs that become training data, managed fine-tuning, and serving for trained adapters. It is currently in alpha and free to use, with paid pricing coming soon.
Key facts
What is Arkor?
One platform for hosting, fine-tuning, and serving open-weight models with full control over weights.
Who is Arkor best for?
Developers and teams building with open-weight models
What are its main limitations?
- Still in alpha, limited availability
- No published pricing for post-alpha
- Currently only supports Gemma 4 (open-weight models)
Limitations are based on product materials or editorial synthesis; confirm on the official site.
Key features
- ✓One-click hosted OpenAI-compatible endpoint with per-deployment API keys
- ✓Request and response logging for inspection and dataset creation
- ✓Managed fine-tuning on GPUs with streaming metrics and checkpoint storage
- ✓Serve trained LoRA adapters on the same endpoint or separate URL
- ✓Rollback and inspect every checkpoint
- ✓Roadmap includes subscription pricing for inference, distillation, and confidential computing
Use cases
- →Swap an OpenAI-compatible endpoint into your existing app without code changes
- →Use production traffic as training data for fine-tuning
- →Serve personalized or task-specific LoRA adapters at scale
- →Iterate on model quality with managed fine-tuning and checkpoint rollback
Pricing
Free (Alpha)
$0
- · Full platform access during alpha
Free during alpha; paid plans coming soon
Pros
- +End-to-end platform from hosting to fine-tuning to serving
- +OpenAI-compatible endpoint simplifies integration
- +Free to use during alpha
- +Open-source framework for fine-tuning (TypeScript)
Cons
- −Still in alpha, limited availability
- −No published pricing for post-alpha
- −Currently only supports Gemma 4 (open-weight models)
Positioning
- Core value: One platform for hosting, fine-tuning, and serving open-weight models with full control over weights.
- Ideal for: Developers and teams building with open-weight models
- Product type: Web app (hosted endpoint) and TypeScript SDK
Frequently asked questions
What is Arkor?
Arkor hosts the entire loop for open-weight models: an OpenAI-compatible endpoint, request logs as training data, managed fine-tuning, and serving for trained adapters.
How much does Arkor cost?
Arkor is currently free during alpha; pricing has not been published yet.
Is there an API?
Yes, Arkor provides an OpenAI-compatible endpoint with per-deployment API keys.
Can I fine-tune models with Arkor?
Yes, Arkor offers managed fine-tuning on GPUs with streaming metrics and checkpoint storage.
What models are supported?
Arkor currently supports Gemma 4 and other open-weight models (determined by the open-source framework).
Is Arkor open-source?
The fine-tuning framework is open-source and written in TypeScript.
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.
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.
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.