News

Announcements from Modellane: new models, price changes and new API features, newest first, with links to the details.

Release notes for new models, price changes and new API features, newest first.

Announcements about new models, price changes and new API features, newest first. For a short dated list of every change, see the Changelog.

Responses and Anthropic Messages formats, and 21 integration guides#

2026-10-03

The API now speaks two more request formats on the same models, keys and prices:

  • OpenAI Responses at POST /v1/responses, on the base URL https://usemodellane.com/v1. This is the format Codex uses, and the one behind client.responses in the OpenAI SDKs. Responses are stateless: send the whole conversation with every request. See the Responses API guide and Responses.
  • Anthropic Messages at POST /v1/messages, plus POST /v1/messages/count_tokens, on the base URL https://usemodellane.com. The Anthropic SDKs and Claude Code work with the base URL, your key and a Modellane model id. See the Anthropic Messages API guide and Messages.

Billing, rate limits, caching and errors work the same way in all three formats; Chat Completions stays the default.

New integration guides#

Integrations now has setup guides for 21 tools, grouped by kind, each with the exact settings, a test prompt and the caveats we know of:

  • Coding agents: Codex, Claude Code, Cline, Roo Code, Kilo Code, Aider, OpenCode, Goose, OpenHands, Continue and Zed.
  • Agent frameworks and assistants: Hermes Agent, OpenClaw, LangChain and LlamaIndex.
  • Chat apps: SillyTavern, Open WebUI and LibreChat.
  • SDKs and libraries: the OpenAI SDK, the Anthropic SDK and the Vercel AI SDK.

The Modellane API is live#

2026-10-03

The Modellane API is open. It follows the OpenAI Chat Completions format, so the SDKs, frameworks and agent tools you already use can switch with two settings: the base URL https://usemodellane.com/v1 and your API key.

What you get#

  • 3 models at launch, each with its context length, output limit, features and per-token price in Models & pricing. The table is read from the live catalog, so it always matches what you are billed.
  • Streaming with server-sent events and an optional usage chunk at the end.
  • Tool calls and JSON output on models that support them.
  • Image input as base64 data URLs on models with Vision.
  • Automatic context caching: repeated prompt prefixes are billed at the lower cached input price, with no code change.
  • Pay as you go: you only pay for the tokens in each response's usage block. A request that fails before returning any content costs nothing.

Get started#

  1. Create a key on the API keys page and add credit on Billing.
  2. Send your first API call.
  3. Try models side by side in the Playground before you write code.

Setup notes for SDKs, coding agents and chat apps are on the Integrations page in the Get started tab.