Inference — Layer 01

Serving, the way agents need it.

Most inference is served for chat: one question, one answer, goodbye. Sideren's engine was built from the ground up for agents — runs that last hours, hammer tools, and cannot afford a dropped call. You point an agent at the endpoint; the engine guarantees the rest.

The guarantees[01/04]

Five guarantees, held on every request.

You point an agent at an endpoint. We do the rest. You never see the machinery — you see what it guarantees.

  1. 01

    Always on

    Degradation never reaches your run — serving recovers before your agent notices, with no retries to write and no fallback lists to maintain on your side. Zero downtime is the design target, not a status-page apology.

  2. 02

    Fast for agents, from the ground up

    Chat inference optimizes one answer. Agent inference is different: hours-long trajectories, tool calls every few seconds, context that keeps growing. Our stack was built for that shape of traffic from day one — sustained throughput across the whole run, not a fast first token and a slow everything else.

  3. 03

    Usage that reads huge

    Plans carry a monthly pool of included usage measured in dollars at public list rates — $45 of model usage on a $5 plan, $200 on Pro. On a paid plan every model in the catalog draws from that one pool: no per-model surcharges, no premium gate, nothing to tune or configure.

  4. 04

    Tool calls that hold

    Agent loops live and die on tool calling. The engine enforces strict, schema-faithful tool calls under load, on every model in the catalog, in both agent dialects — so a malformed call never silently derails a run.

  5. 05

    Any agent, no ceremony

    Anthropic-compatible and OpenAI-compatible endpoints, streaming everywhere, one key. If it runs an agent loop, it runs here — the engine does not care which framework is driving.

Point an agent here →
REQUEST PATHCUTAWAY
The catalog[03/04]

Every model, behind one endpoint.

You hold one key and one endpoint — never another vendor account, quota, or bill. Open-weight flagships sit next to premium frontier models behind the same door, and the catalog keeps growing. One key reaches all of it.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.sideren.io/v1",
    api_key="sdn_your_key",
)

response = client.chat.completions.create(
    model="sdn-gpt-oss-120b",
    messages=[{"role": "user", "content": "Say hello."}],
)
Both dialects · one sdn_ key · every model in the catalogFull API reference →
Live in the catalog todaygrowing
Claude Opus 5frontierClaude Opus 4.7frontierGPT-5.5frontierGPT-5.4frontierGemini 3 ProfrontierKimi K3frontierDeepSeek V4 ProfrontierQwen3.7 MaxfrontierInklingfrontierGPT-OSS 120Bgeneral purposeSonnet 4.6premiumClaude Sonnet 5premiumClaude Haiku 4.5premiumGrok 4.3premiumGLM 5.2codingGLM 5codingQwen3 Coder 480BcodingKimi K2.7 CodecodingQwen3-MaxcodingQwen3 235BreasoningDeepSeek V3.2reasoningKimi K2 ThinkingreasoningMiniMax M2.5reasoningERNIE X1reasoningHunyuan T1reasoningStep 3reasoningQwen3.5 397BreasoningNemotron 3 UltrareasoningNemotron Super 3 120Bgeneral purposeNemotron 3 120Bgeneral purposeQwen3 Next 80Bgeneral purposeGemma 4 31Bgeneral purposeMistral Large 3 675Bgeneral purposeERNIE 5.1general purpose
Hunyuan TurboSgeneral purposeGemini 3 Flashgeneral purposeMiniMax M3general purposeKimi K2.6general purposeGLM 4.7general purposeQwen3.7 Plusgeneral purposeQwen3 VL 235Bgeneral purposeHunyuan Hy3general purposeLlama 3.3 70Bgeneral purposeQwen3 Coder 30Beveryday codingQwen3 Coder Nexteveryday codingKimi K2.5everyday codingDevstral 2 123Beveryday codingMiniMax M2.7everyday codingKAT Coder Proeveryday codingGemma 4 26Blong contextLongCat 2.0codingDeepSeek V4 Flashlong contextQwen3.5 Flashlong contextMiMo V2.5long contextQwen TurbofastQwen 3.8 MaxreasoningGPT-OSS 20BfastGLM 4.7 FlashfastNemotron Nano 3 30BfastStep 3.7 FlashfastBGE-M3embeddingsBGE Base EN v1.5embeddingsBGE Small EN v1.5embeddingsBGE Large EN v1.5embeddingsEmbeddingGemma 300MembeddingsQwen3 Embedding 0.6BembeddingsPLaMo Embedding 1Bembeddings
Browse every model →
Works under your agent[04/04]

Point any agent here in one command.

Run npx sideren and pick your tools — the installer writes each config where the tool expects it, key and model included; GUI-configured tools get their exact paste-in values. No SDK, no code changes. The session below is the setup.

npx siderenlive session · replay
$ npx sideren Sideren · configure your coding agent Press enter to accept the [default]. Gateway URL: [https://api.sideren.io] API key (sdn_…): sdn_•••••••••••• Default model: [sdn-gpt-oss-120b] Which tools? (comma-separated numbers, or 'all') 1. Claude Code · Anthropic /v1/messages 2. OpenAI Codex CLI · OpenAI Responses /v1/responses 3. Aider · OpenAI /v1/chat/completions (via LiteLLM) 4. Continue (VS Code / JetBrains) · OpenAI /v1/chat/completions 5. Cline (VS Code) · OpenAI /v1/chat/completions 6. Cursor · OpenAI /v1/chat/completions 7. Windsurf · OpenAI /v1/chat/completions Selection: [all] 1,2 Configuring → https://api.sideren.io model=sdn-gpt-oss-120b ✓ Claude Code → wrote ~/.claude/settings.json • If you previously logged in to Claude, run `/logout` once so the gateway key is used. • Then just run `claude`. ✓ OpenAI Codex CLI → wrote ~/.codex/config.toml • Persist your key: export SIDEREN_API_KEY in your shell profile (setup offers to do this for you). • Codex requires the Responses API — Sideren serves /v1/responses. • Then run `codex`. Persist SIDEREN_API_KEY to your shell profile now? (Y/n) [Y] ✓ SIDEREN_API_KEY added to ~/.zshrc — restart your shell (or source it) to pick it up. Done. Run `npx sideren test` to verify the gateway.
Per-tool setup guides →
Put your agent on real infrastructure

Your agent doesn't change.
Everything underneath does.

$export ANTHROPIC_BASE_URL=https://api.sideren.io

Start free · no card · Starter from $5/mo