Together AI
Hosted platform to run, train and serve open-source models via prepaid serverless, batch, dedicated and fine-tuning APIs, with an OpenAI-compatible base URL.
Tags
Description
Commercial Together AI platform (together.ai). Established provider gap fill, not a claim of a brand-new launch. https://together.ai 301s to https://www.together.ai/; use the www host. https://docs.together.ai/ and https://docs.together.ai both 308 to https://docs.together.ai/intro (HTTP 200), so the docs URL is the intro page, not the docs root. Official overview: serverless inference is charged on tokens used and model size; dedicated inference is charged on GPU hours. The live pricing table at https://www.together.ai/pricing (200) is authoritative (per-1M-token serverless rates; some rows are listed as Free; also batch, provisioned throughput, fine-tuning and GPU clusters). Individual rates are not copied here because the table changes. Billing docs (https://docs.together.ai/docs/billing-credits, 200): Together is fully prepaid and does not currently offer a free trial. The first credit purchase must be at least $15 (free credits do not count toward that minimum); later purchases have a $5 minimum. Until that first purchase, the console does not allow playground use, API key creation, or inviting teammates. A positive purchased-credit balance is required to keep API access; if the balance reaches zero, access is suspended until credits are added. Prepaid balance credits do not currently expire. Auto-recharge can charge a default card when the balance falls below a threshold. Scale/Enterprise contracts are billed under contract terms instead. Refund terms were not independently extracted beyond that credits page. No purchase was made. Auth is a project-scoped API key (TOGETHER_API_KEY; keys can spend the project's credit balance, including via auto-recharge; not created). Official Python (pip install together / uv add together) and TypeScript (npm install together-ai) SDKs, or the OpenAI SDK pointed at Together. OpenAI compatibility (https://docs.together.ai/docs/inference/openai-compatibility, 200): set the OpenAI client base URL to https://api.together.ai/v1 and the API key to a Together key. Together also documents the OpenAI Responses API and the Anthropic Messages API. A US-only base URL https://us-api.together.ai/v1 is documented to charge 1.25× standard serverless rates for input, cached input, and output tokens. Assistants/threads/runs, OpenAI-shaped fine-tuning and batches, and moderations.create are documented as not supported (Together-native APIs exist for some of those). OpenAPI spec https://docs.together.ai/openapi.yaml and llms.txt https://docs.together.ai/llms.txt were HTTP 200. license proprietary; pricing paid. No signup or spend was used.
Agent access
Related
- OpenRouter: Unified API for hundreds of LLMs from many providers behind one endpoint and key, with model and price comparison.
- Fireworks AI: Hosted inference and training platform for open models: prepaid pay-per-token serverless, reserved throughput, dedicated GPUs and fine-tuning, with OpenAI- and Anthropic-compatible APIs.
Guides & comparisons
- GPU clouds and serverless inference for agents (Comparison): Compare GPU rental, serverless GPUs and hosted per-token APIs for agents: OpenAI-compatible base URLs, tools and structured output, BYO models, scale-to-zero and billing, APIs/CLIs, and host-trust notes. No cached dollar figures or invented benchmarks.
Sources
- https://www.together.ai
- https://www.together.ai/pricing
- https://docs.together.ai/intro
- https://docs.together.ai/llms.txt
- https://docs.together.ai/docs/quickstart
- https://docs.together.ai/docs/inference/openai-compatibility
- https://docs.together.ai/docs/api-keys-authentication
- https://docs.together.ai/docs/billing-credits
- https://docs.together.ai/openapi.yaml
- https://github.com/togethercomputer/together-python
Machine-readable
- JSON
- Markdown
- Source file on GitHub (edit via pull request)