Gemini API support, new models, streaming, and public status page¶
This major Nexus API release significantly broadens the ecosystem of supported protocols: introducing native Gemini API compatibility, new GPT-5.6 and Gemini 3 model lines, streaming for Chat Completions, connection stability standards, and a public service status page.
New capabilities¶
Native Gemini API support and model family¶
- Added native Gemini API endpoints: text generation
(
POST /v1beta/models/{model}:generateContent), stream generation (POST /v1beta/models/{model}:streamGenerateContent), token counting (POST /v1beta/models/{model}:countTokens), and model discovery viaGET /v1beta/modelsandGET /v1beta/models/{model}. - Added the Gemini model family to the catalog and Telegram bot: Gemini 3 Flash, Gemini 3.5 Flash, Gemini 3.1 Pro Preview, and Gemini 3.1 Flash Image Preview.
- Grouped the Gemini family in the Models menu and key permission settings.
New OpenAI-compatible models and streaming¶
- Added next-generation flagship models: GPT-5.6 Luna, GPT-5.6 Terra, and GPT-5.6 Sol.
- Implemented real-time Server-Sent Events (SSE) streaming support for
POST /v1/chat/completions, enabling token-by-token streaming responses. - Added support for Legacy Completions (
POST /v1/completions) and compact chat aliases for integration with older development tools. - Added a direct single-model lookup endpoint:
GET /v1/models/{model_id}.
Public service status page¶
- Launched the official status page at status.nexus-hub.tech monitoring latency and availability across all supported endpoints.
- Added a direct button in the Telegram bot main menu to inspect service health.
Connection stability and standardized error handling¶
- Established long-running connection policies: graceful handling of pauses between streaming chunks to prevent false timeouts.
- Safe client disconnect handling: early connection drops do not leave dangling resources or trigger incorrect balance deductions.
- Standardized JSON error response envelopes with a unique
request_idfor efficient support diagnostics.
Improvements¶
Price reductions and bonus credits¶
- Implemented an across-the-board 5–15% base rate reduction for major models.
- Automatically credited bonus balances to users registered prior to the release.
- Streamlined bot price tables: hidden unpriced companion endpoints and consolidated multi-endpoint models into clean single rows.
Developer tooling compatibility¶
- Improved compatibility for developer tools: Codex CLI, Claude Code, VS Code, Zed, JetBrains, OpenCode, and custom AI agents.
Fixed issues¶
API and protocols¶
- Fixed empty responses and format discrepancies in Legacy Completions and compact model responses.
- Stopped forwarding client
User-Agentheaders to preserve client environment privacy. - Fixed token usage reporting in Chat Completions streaming mode.
Catalog and bot interface¶
- Fixed non-deterministic model sorting during concurrent catalog requests.
- Fixed keyboard button bindings for the status page in the Telegram bot.
Compatibility and user actions¶
- Existing API keys and base URLs remain fully compatible with all new features; no key reissuance is needed.
- To connect tools to the Gemini API, use the standard
/v1beta/base URLs and endpoints with your Nexus Hub API key. - Review the API reference and client setup guides for configuration details.