Changelog
Shipped, not promised.
What ships, as it ships — dates in UTC.
Changelog — all entriesUTC
2026·10·23Better cache hits in long sessions — Requests from the same agent session now go back to the same warm cache, so more of your repeated context is served from cache and uses fewer credits.
2026·10·22Web search for your agents — npx sideren setup now adds Sideren Search to Claude Code, Codex and OpenCode. Search and page fetch run from your own machine and need no extra key.
2026·10·22Model lists that stay current — Agents set up with npx sideren refresh their Sideren model list on their own when the catalog or your plan changes, so you don't need to run setup again.
2026·10·21Compaction — Long conversations can now compact. Older turns are replaced with a summary and the run keeps going. Use the OpenAI or Anthropic compaction features you already know, or turn it on for chat completions with one header.
2026·10·21DeepSeek V4.1 Flash 1017 — Now in the catalog on every plan, Free included.
2026·10·20MiniMax M3.1 Flash — MiniMax's newest coding model, with a million-token window, is live on every paid plan.
2026·10·19Sarvam 105B — Live on every paid plan. Sarvam 105B Conversations is on Free.
2026·10·19Docs at docs.sideren.io — The documentation has moved to docs.sideren.io. Old links redirect.
2026·10·18A new Free lineup — Free now runs open models from DeepSeek, Kimi, MiniMax, Qwen, Mistral and others, with 200 one-time credits to try them.
2026·10·17New dashboard and auto top-up — The dashboard is now black and white and easier to read. Save a card and auto top-up refills your purchased credit before it runs out.
2026·10·08Smoother streaming — Text now streams in steady, word-sized steps instead of bursts.
2026·10·07Public model ids and credits — Models have stable public ids such as @sdn/glm-5.3, and the dashboard shows usage in credits.
2026·10·01Monthly credits on every plan — Every paid plan now comes with monthly credits: Starter $5, Pro $19, Max $49, and the new Scale plan at $99.
2026·09·30LayerX1 is now Sideren — Same API and same keys under a new name, at sideren.io. Old addresses redirect.
2026·09·28Claude Sonnet 5.5 — Anthropic's newest Sonnet is live on Max and Scale.
2026·09·05Launch plan — $2/month now includes $30 of model usage, 100 requests every 5 minutes, and five simultaneous runs — live from Billing on the key you already have.
2026·07·07Setup guides for 12 more agent harnesses — opencode, Crush, Goose, Hermes, OpenClaw, Qwen Code, OpenHands, Zed, Aider, Continue, Cline and Kilo Code — each with verified, copy-paste configuration in the docs.
2026·07·06Serving engine v2 — Higher burst headroom and steadier serving under peak load, across every model in the catalog.
2026·07·04Streams that survive — If a stream dies before the first token arrives, the engine fails over automatically — your agent never sees the gap.
2026·07·04Paid plans open — Starter $5, Pro $19 and Max $49 subscriptions are live — upgrade from the dashboard, effective immediately.
2026·07·03api.sideren.io — The gateway moved to its canonical endpoint, with Google sign-in and name-on-signup joining the account flow.
2026·07·03Embeddings and legacy completions — /v1/embeddings and /v1/completions join messages, chat completions and responses — four protocols on one endpoint.
2026·07·03Open signup + one-command setup — Accounts are open to everyone, with a welcome onboarding — and npx sideren configures your coding agent in one command.
2026·07·02Usage dashboard — Live usage meters, per-request logs, and your plan's included usage — all visible in the dashboard as they happen.
2026·07·01Image input — Vision-capable models now accept images in both API dialects, including screenshots pasted mid-run by coding agents.
Put your agent on inference built for the workYour agent stays the same.
Your agent stays the same.
Its inference gets better.
$export ANTHROPIC_BASE_URL=https://api.sideren.io
Start free · no card · Paid plans from $5/mo