# Tarka

> Provision the components your agents need, from inference and hosting to sandboxes and storage, from the CLI or your coding agent over MCP, on one account with one bill.

These Markdown pages mirror the public website and are regenerated with every build.
Each page includes its original website URL. Product availability and API examples are documented on the linked pages.

## Tarka

- [About Tarka — AI and agent infrastructure from Nepal](https://tarkahq.com/about/index.md): Tarka is an AI and agent infrastructure company based in Kathmandu, building model APIs and runtimes for developers and organizations.
- [AI for Nepal — Tarka](https://tarkahq.com/ai-for-nepal/index.md): HimalayaAI's Nepali language, OCR and speech models on one OpenAI-compatible API, and hosting on Tarka for other models built for Nepal.
- [Enterprise AI infrastructure — Tarka](https://tarkahq.com/enterprise/index.md): Private model routes, dedicated AI capacity, deployment review, and direct engineering support from Tarka.
- [How Tarka works: your coding agent sets it up, Tarka runs it](https://tarkahq.com/how-it-works/index.md): How your coding agent sets up an agent on Tarka, how your product calls it, and how the blocks fit together for one agent and for a whole application.
- [Tarka — Build and deploy AI agents with ease](https://tarkahq.com/index.md): Provision the components your agents need, from inference and hosting to sandboxes and storage, from the CLI or your coding agent over MCP, on one account with one bill.
- [Tarka Provider Network](https://tarkahq.com/network/index.md): Offer models or GPU capacity to Tarka customers through one managed gateway.

## Documentation

- [Batch Jobs \| Tarka Docs](https://tarkahq.com/docs/compute/batch-jobs/index.md): Availability, queued-run contracts, and billing consent for Tarka Batch Jobs.
- [Hosted Agents \| Tarka Docs](https://tarkahq.com/docs/compute/hosted-agents/index.md): Deploy complete Compose-defined agent stacks with prebuilt images, state, networking, HTTPS, and tenant isolation.
- [Compute \| Tarka Docs](https://tarkahq.com/docs/compute/index.md): Documentation for Tarka Hosted Agents, Batch Jobs, and Sandboxes.
- [Browser Sandboxes \| Tarka Docs](https://tarkahq.com/docs/compute/sandboxes/browser/index.md): Documentation placeholder for Tarka Browser Sandboxes.
- [Code Sandboxes API \| Tarka Docs](https://tarkahq.com/docs/compute/sandboxes/code/index.md): Create dependency templates, start isolated Python or Node.js sandboxes, and execute source code with the Tarka API.
- [Sandboxes \| Tarka Docs](https://tarkahq.com/docs/compute/sandboxes/index.md): Browse Tarka Code, Linux, and Browser Sandbox documentation.
- [Linux Sandboxes \| Tarka Docs](https://tarkahq.com/docs/compute/sandboxes/linux/index.md): Documentation placeholder for Tarka Linux Sandboxes.
- [Control API authentication \| Tarka Docs](https://tarkahq.com/docs/control-api/authentication/index.md): Authenticate Tarka Control API REST and gRPC requests and understand account, membership, role, entitlement, and API-key boundaries.
- [Billing and compute quotes \| Tarka Docs](https://tarkahq.com/docs/control-api/billing-quotes/index.md): Review compute prices, prepaid reservations, usage policy consent, retention and billing errors.
- [Control API Object Storage and Hosted Git \| Tarka Docs](https://tarkahq.com/docs/control-api/data/index.md): Provision Tarka Object Storage buckets, scoped S3 credentials, Hosted Git identities, and private repositories through the Control API.
- [Control API \| Tarka Docs](https://tarkahq.com/docs/control-api/index.md): Automate Tarka organizations, credentials, policy, usage, compute, storage, Git, and sandboxes through the Control API.
- [Inference Capture \| Tarka Control API Docs](https://tarkahq.com/docs/control-api/inference-capture/index.md): Configure organization inference capture through the console, Control API, or MCP, and understand stored records, media, expiry, and delivery health.
- [Control API keys, policy, and usage \| Tarka Docs](https://tarkahq.com/docs/control-api/keys-policy-usage/index.md): Discover active Tarka models, manage scoped API keys, configure organization limits, and inspect usage through the Control API.
- [Connect Claude and MCP clients \| Tarka Docs](https://tarkahq.com/docs/control-api/mcp/index.md): Connect Claude, Claude Code, or another MCP client to Tarka with OAuth, and see what read and write access allow, which tools exist, and how to end a connection.
- [Control API accounts and organizations \| Tarka Docs](https://tarkahq.com/docs/control-api/organizations/index.md): Bootstrap and manage Tarka organizations, list members, and request access to products.
- [Control API provisioned resources \| Tarka Docs](https://tarkahq.com/docs/control-api/provisioning/index.md): Apply and inspect desired state for managed Tarka inference services, Agent Hosts, and Batch Jobs.
- [Control API Code Sandboxes \| Tarka Docs](https://tarkahq.com/docs/control-api/sandboxes/index.md): Manage Tarka Code Sandbox templates, instances, execution, idempotency, and output retention through the Control API.
- [Data residency \| Tarka Docs](https://tarkahq.com/docs/data-residency/index.md): Where Tarka compute, storage and model processing run, including unverified locations.
- [Hosted Git \| Tarka Docs](https://tarkahq.com/docs/data/hosted-git/index.md): Create private Tarka Hosted Git repositories and use SSH, HTTPS, and Git LFS.
- [Data products \| Tarka Docs](https://tarkahq.com/docs/data/index.md): Documentation for Tarka Object Storage and Hosted Git.
- [Object Storage \| Tarka Docs](https://tarkahq.com/docs/data/object-storage/index.md): Create private S3-compatible buckets and scoped credentials with Tarka Object Storage.
- [Tarka Documentation](https://tarkahq.com/docs/index.md): Documentation for Tarka inference, compute, storage, and Hosted Git products.
- [Inference API access \| Tarka Docs](https://tarkahq.com/docs/inference/api-access/index.md): Choose a Tarka inference endpoint, authenticate public requests, discover models, and handle API errors.
- [Inference \| Tarka Docs](https://tarkahq.com/docs/inference/index.md): Integrate Tarka language and utility models through versioned inference APIs.
- [Language model inference \| Tarka Docs](https://tarkahq.com/docs/inference/llms/index.md): Create OpenAI-compatible Chat Completions and Responses with Tarka language models.
- [Utility model APIs \| Tarka Docs](https://tarkahq.com/docs/inference/utility-models/index.md): Use Tarka utility models through canonical v1 REST and native gRPC.

## Products and models

- [Tarka Compute: hosted agents, batch jobs and code sandboxes](https://tarkahq.com/compute/index.md): Keep agents online, run batch jobs to completion and execute generated code in isolated sandboxes, all on one Tarka account.
- [Tarka Data: object storage and hosted Git](https://tarkahq.com/data/index.md): Private S3-compatible object storage and hosted Git repositories in the same Tarka account as your agents and models.
- [DeepSeek V4.1 Flash \| Tarka Inference](https://tarkahq.com/inference/deepseek-v4-1-flash/index.md): DeepSeek's fast open model for everyday chat, tool use, and coding.
- [GLM-5.3 Flash \| Tarka Inference](https://tarkahq.com/inference/glm-5-3-flash/index.md): Z.ai's efficient open model for high-volume chat and extraction.
- [GLM 5.3 \| Tarka Inference](https://tarkahq.com/inference/glm-5-3/index.md): GLM 5.3 through Tarka's OpenAI-compatible API.
- [Himalaya Gemma 4 E2B \| Tarka Inference](https://tarkahq.com/inference/himalaya-gemma-4/index.md): A bilingual Nepali–English Gemma 4 fine-tune from HimalayaAI.
- [HimalayaGPT 0.5B \| Tarka Inference](https://tarkahq.com/inference/himalayagpt-0-5b/index.md): Sovereign Nepali LLM by Himalaya AI Research Lab, hosted in Kathmandu.
- [OpenAI-compatible model inference for agents and apps \| Tarka](https://tarkahq.com/inference/index.md): Call GLM-5.3 Flash, DeepSeek V4.1 Flash, Kimi K3 and speech, OCR, and voice models through Tarka’s OpenAI-compatible API.
- [Kimi K3 \| Tarka Inference](https://tarkahq.com/inference/kimi-k3/index.md): Moonshot AI's flagship open model, for coding agents and long, multi-step tasks.
- [Qwen 3.8 Flash Next \| Tarka Inference](https://tarkahq.com/inference/qwen-3-8-flash-next/index.md): A free, self-hosted language model through Tarka's API.
- [Pricing — Tarka](https://tarkahq.com/pricing/index.md): Price an agent on Tarka or in your own infrastructure, and see the unit prices for model APIs and managed compute, billed from prepaid credit.
- [Code Runtime \| Tarka Sandboxes](https://tarkahq.com/sandboxes/code-runtime/index.md): Python and Node sandboxes for short evals, code generation, and tool use.
- [GLM-OCR \| Tarka Utilities](https://tarkahq.com/utilities/glm-ocr/index.md): Document vision for receipts, PDFs, screenshots, forms.
- [Kokoro \| Tarka Utilities](https://tarkahq.com/utilities/kokoro/index.md): Small, fast TTS — built for low-latency agent loops.
- [Nepali Parler TTS \| Tarka Utilities](https://tarkahq.com/utilities/nepali-parler-tts/index.md): Nepali text to speech with natural-language delivery instructions.
- [PaddleOCR-VL 1.6 \| Tarka Utilities](https://tarkahq.com/utilities/paddleocr-vl-1-6/index.md): Compact multilingual document OCR for text, tables, formulas, charts, and seals.
- [Piper \| Tarka Utilities](https://tarkahq.com/utilities/piper/index.md): CPU-friendly English TTS with a pinned, predictable voice.
- [PP-DocLayoutV3 \| Tarka Utilities](https://tarkahq.com/utilities/pp-doclayout-v3/index.md): Layout analysis for text, titles, tables, formulas, figures, charts, and curved pages.
- [Qwen3-ASR \| Tarka Utilities](https://tarkahq.com/utilities/qwen3-asr/index.md): Apache 2.0 speech recognition with Qwen3-ASR 1.7B.
- [Qwen3-TTS \| Tarka Utilities](https://tarkahq.com/utilities/qwen3-tts/index.md): Apache 2.0 text to speech with consent-backed voice cloning.
- [Whisper Large v3 \| Tarka Utilities](https://tarkahq.com/utilities/whisper-large-v3/index.md): Multilingual speech-to-text — 99 languages, including Nepali.
- [Whisper Medium Nepali \| Tarka Utilities](https://tarkahq.com/utilities/whisper-nepali-medium/index.md): A Nepali-specialized Whisper Medium route for direct transcription.

## Legal

- [Data handling and zero data retention — Tarka](https://tarkahq.com/data-handling/index.md): What Tarka keeps and what it discards: zero data retention on Tarka-served models by default, opt-in capture into your own bucket, partner-served and bring-your-own-key models, and the metadata Tarka keeps to run and improve the platform.
- [Imprint — Tarka](https://tarkahq.com/imprint/index.md): The legal entities behind tarkahq.com: TarkaHQ, Inc., a Delaware corporation, and its subsidiary Tarka Pvt. Ltd. in Kathmandu, Nepal.
- [Privacy Policy — Tarka](https://tarkahq.com/privacy/index.md): How Tarka handles personal data and customer data: zero data retention on Tarka-served models by default, what is kept for billing and operations, subprocessors, cross-border transfers and your rights.
- [Terms of Service — Tarka](https://tarkahq.com/terms/index.md): The terms for using Tarka's inference, compute and data products: accounts, acceptable use, your content and ours, fees and prepaid credit, availability, liability and termination.