Changelog

What’s New

New features, improvements and fixes to the SovereignEG platform, newest first.

betaImprovement

Rerank models on the pricing page

The pricing page now lists the rerank models in a group of their own, each at its price per 1M input tokens (the query plus the documents) or per search. A rerank request is billed once and has no output charge.

betaImprovement

Clearer rerank and embedding model pages

A rerank model's page now says how to call it, over HTTP with a POST request to /v1/rerank, and links to the Rerank guide; its FAQ gives the price of one request. Rerank and embedding model pages no longer show a maximum output or a monthly cost example in output tokens: these models return relevance scores or vectors, not generated text, and are billed on their input. In the model catalog, their cards no longer have a Try it link, since the Playground runs chat and image models only.

betaFix

A too-large rerank request says to send fewer or smaller documents

A POST /v1/rerank request whose body is over the 32 MiB limit gets 413 request_too_large, and its message now ends with "Send fewer or smaller documents." instead of the advice for chat requests about messages, tool outputs and attachments. The message on every other endpoint is unchanged. The Error Codes and Rate Limits pages of the docs say the same.

betaFix

Arabic pricing page links, and image prices in the Arabic dashboard

On the Arabic pricing page, a model name now opens the model's Arabic page instead of the English one. In the dashboard in Arabic, an image model's Pricing tab now shows its price in Arabic.

betaFeature

Rerank models on /v1/rerank

POST /v1/rerank ranks a list of documents by how relevant each one is to a query: use it after a fast first search (keyword or embeddings) to put the best results on top. It takes the same request as OpenRouter's rerank API, up to 1,000 documents at a time, and returns the results sorted by relevance; call it over HTTP with your usual API key. Each request is billed once at the model's catalog price in EGP, per 1M input tokens or per search, and usage.cost on the response is that price. Rerank models are listed in the model catalog under Modality: Rerank.

betaFix

Function tools on gpt-5.4, gpt-5.5, gpt-6.1-sol and gpt-6-astra

A request with function tools no longer fails with a 400 on these OpenAI models. On gpt-5.4, gpt-5.4-mini, gpt-5.4-nano and gpt-5.5, a request with function tools and a reasoning_effort other than none is sent to OpenAI with none, the only value OpenAI takes together with function tools on chat completions, and the X-Mapped-Params response header says so; a request that sends no reasoning_effort goes out as before. gpt-6.1-sol and gpt-6-astra refuse none as well, so OpenAI takes function tools with them only through its Responses API: such a request is now served by OpenRouter (with the openai/ prefix, which keeps a request on OpenAI, it gets a 400 on tools that says so). Neither model can turn reasoning off, so a none you send to them goes out as low when OpenAI serves the request and as minimal when OpenRouter does, and X-Mapped-Params says so.

betaFix

A declined Claude request shows why, and is usually free

When Anthropic's safety filter declines a Claude request, the answer now carries its reason in refusal (delta.refusal when streaming) next to finish_reason content_filter, and an answer with no text shows provider_refusal in your request log. A refusal that comes before the model generates anything is now free, except in the categories Anthropic bills (bio, frontier_llm, reasoning_extraction). This applies while Anthropic serves the request. The Playground now says the request was declined and no longer sends the declined message again with your next one.

betaFix

A balance that was never topped up no longer goes below zero

An organization that has never made a paid top-up, and that we do not bill by separate arrangement, is now charged at most its remaining balance: when a request costs more, only the remaining balance is deducted, the difference is not collected later, and the request log shows the amount with "Remaining balance used". Until its first paid top-up, its requests are also sized so that the balance covers them before they start: a non-streaming request holds the most it can cost on every model and is checked against the credit that the requests still running hold, so two requests can no longer spend the same credit (the second is served with a lower max_tokens, or gets 429 inflight_busy while the first still holds the credit); a prompt is held at twice its estimated size; and a request with audio, a file or a video is held at the model's whole context window, so it can get 402 insufficient_balance where a top-up would let it run. Requests on your plan's free daily allowance model are held as before, and so is every request of an organization with a paid top-up, which keeps its small overdraft.

betaFix

Clear errors for a wrong base URL and for plan limits

A request to a path that clients commonly get wrong now answers 404 route_not_found with the fix and the exact base URL in the message: /v1 written twice, /v1/messages (Claude Code and the Anthropic SDKs cannot connect; use an OpenAI-compatible client such as OpenCode), /v1/responses, /v1/completions, or a base URL without /v1 or ending at /chat/completions. On the Student plan and the Developer Pack, a refusal for the prompt size, the requests per minute or the requests at a time now says that it is the plan's limit, not the model's, and names the plan and the model.

betaImprovement

OpenCode guide in Arabic

The OpenCode guide now gives the exact base URL, the setting for plans with a free daily allowance (limit.context a little under the plan's prompt cap, so OpenCode shortens the conversation before it reaches the cap) and the common errors with their fixes. It is now also available in Arabic.

betaImprovement

Tool names up to 256 characters

A tool name can now be up to 256 characters instead of 64, so the long <server>_<tool> names that MCP clients such as OpenCode build no longer fail. Where a model's provider takes only shorter names (64 characters on OpenAI-compatible providers, 128 on Anthropic, and no dot or colon on some of them), it receives a short, stable alias, and the tool calls in the response, streamed or not, carry your own tool names: nothing changes in your code. The alias can still show up where the model writes it itself (its answer, its reasoning or a call's arguments) and in an error message from the provider.

betaFix

gpt-6 with function tools and no reasoning_effort

On gpt-5.6, gpt-6 and later chat models served by OpenAI (not gpt-6-astra, the -pro or the chat-latest models), a request with function tools that sends no reasoning_effort now goes out with reasoning_effort: none too, instead of failing with a 400: these models reason by default, and OpenAI does not take function tools together with reasoning on chat completions. The X-Mapped-Params response header says so.

betaFix

Retry shortly instead of out of credit

A chat completion naming its model that is held up only by your organization's own requests still running now gets 429 inflight_busy with Retry-After: 10 instead of a 402 that read as out of credit, with stream: false too: a streaming request after waiting up to about 25 seconds for them, a non-streaming one at once or after up to about 10. Retry it, as the OpenAI SDKs do by themselves. There a 402 still means the credit is missing even with nothing running. A model: "auto" request can get either: the 402 when what the running requests leave is too little for its classification step or for every model in its auto pool, the 429 when the chosen model's gate refuses it. Embeddings, images and decision models still answer that case with a 402; its message counts the credit held by requests still running (with paying from the balance on, daily_allowance_exhausted says the allowance is used up instead), and a retry once they finish can succeed. A stream refused before its first token because its long prompt reaches a long-prompt rate your credit can't pay (on Claude, for example) is now billed for that prompt, which the provider has already read, only with a 402, never with a 429, and never for more than your credit, or your spend cap, still had room for. If the credit still can't pay it, the same prompt sent again within 10 minutes is refused before it reaches the provider.

betaFix

gpt-6 with max_tokens and tools, and system messages on our own models

Requests to gpt-6 and later GPT models served by OpenAI no longer fail on max_tokens: it is sent as max_completion_tokens, the way OpenAI asks for these models, and temperature, top_p and the penalties, which they do not accept, are left out, as for GPT-5. On gpt-5.6, gpt-6 and later chat models (not gpt-6-astra, the -pro or the chat-latest models), a request with function tools and a reasoning_effort other than none is sent to OpenAI with reasoning_effort: none, the only value OpenAI takes together with function tools on chat completions, instead of failing with a 400, and the X-Mapped-Params response header says so. On the sovereign models on our own GPUs (for example SovereignEG/Qwen3.8-27B-FP8), a second system message, or one later in the conversation, no longer fails with a 400: the system messages at the start are sent as one, and a later one is sent in its place as a user message that starts with [system].

betaImprovement

Key and project spend caps are hard limits

A monthly spend cap on an API key or a project used to stop new requests only: requests already running when the cap was reached finished and were billed in full, so spend could end above the cap. Spend now never goes above a cap you set. A request the cap can only partly pay for is served with max_tokens lowered to what the cap has left (X-Output-Clamped), a stream that reaches the cap ends cleanly with stop_reason: spend_cap_reached, and nothing is billed past it. If the cap cannot be checked for a moment, a request on a capped key or project gets 503 spend_cap_unavailable with Retry-After: 5.

betaImprovement

Claude thinking settings now apply after a failover to OpenRouter too

When Anthropic is overloaded or failing, a Claude request can be served by OpenRouter. OpenRouter now gets the same thinking settings as Anthropic: none switches thinking off on the models that can stop thinking, so a small max_tokens goes to the answer, and minimal to xhigh think at that effort. With no reasoning_effort nothing is sent, as on Anthropic, and Claude 5.x models think at their own default. Sonnet 5.5 cannot stop thinking on OpenRouter, so it thinks at the lowest effort, reported as X-Mapped-Params: reasoning_effort=low. On Claude 4.x and older models reasoning_effort is now removed there too (X-Dropped-Params), as when Anthropic serves the request. If OpenRouter refuses the thinking setting, the request is sent to it once more with reasoning_effort as sent.

betaFix

temperature no longer fails requests to Claude Opus 4.7 and later

Claude Opus 4.7 and later Claude models (Opus 4.8, Opus 5 and 5.5, Sonnet 5 and 5.5, Haiku 5.5, Fable) accept only their default sampling, so a request that set temperature to anything but 1, or set top_p, got a 400 from Anthropic ("`temperature` is deprecated for this model."). While Anthropic serves these models, the gateway now leaves temperature and top_p out of the request, the model samples at its defaults, and the X-Dropped-Params response header names what was left out. Claude Opus 4.6, Sonnet 4.6, Haiku 4.5 and older models still receive the value you set.

betaImprovement

A streaming request on a low balance is served instead of refused

A streaming request your credit could not fund to 512 output tokens used to get 402 insufficient_balance, while the same request with stream: false was served. It is now served the same way, with max_tokens lowered to what your credit buys (X-Output-Clamped). On models that reason before they answer where you cannot see it (Claude, OpenAI GPT-5 and o-series, Gemini thinking models, Grok) the cap is never lowered below 1,024 output tokens, because the hidden reasoning could use it all up and leave an empty answer. A request held up only by your organization's other running requests now waits up to about 25 seconds for them, key and project spend caps included, before 429 inflight_busy. A few models measured ignoring max_tokens now end with stop_reason: cap_exceeded, and output past max_tokens is not billed.

betaImprovement

reasoning_effort now controls Claude thinking

On Claude models served directly by Anthropic, reasoning_effort now sets thinking: none turns it off on the models that can stop thinking (Haiku 5.5, Sonnet 5.5, Sonnet 5, Opus 5), and minimal to xhigh think at that effort. Opus 5.5 and the Fable models cannot stop thinking, so they think at the lowest effort, reported as X-Mapped-Params: reasoning_effort=low. Thinking counts toward max_tokens and is billed as output, including thinking that is not shown, and Claude models newer than the gateway's table think by default with the thinking hidden.

betaFix

reasoning_effort high is accepted on the sovereign Qwen3.8 model

SovereignEG/Qwen3.8-27B-FP8 runs on our own GPUs, and its chat template accepts only low, medium and xhigh for reasoning_effort, so a request that sent high (or minimal) returned the engine 400. The gateway now maps the value to the nearest one the model accepts (high becomes xhigh, minimal becomes low) and names it in the X-Mapped-Params response header, the same adjustment it already made for OpenAI reasoning models. none still goes through unchanged and turns reasoning off.

betaFix

A streaming request that fails before the stream starts returns a real HTTP error

A stream: true request that failed before the stream started (a provider rejecting the request, no healthy backend, an empty response) used to return 200 OK with the error inside the stream, so SDKs and agent frameworks read it as success and then failed on the error event. It now returns the same HTTP status and JSON error body as a non-streaming request, and the OpenAI SDKs raise it from the create call. A provider that reports an error as its first stream event is treated the same way, and a retriable one is retried on another backing. Errors after the stream has started still arrive inside the stream. The X-Dropped-Params / X-Mapped-Params headers on a stream now describe the backing that served it, as they already did on non-streaming responses.

betaFix

reasoning_effort no longer fails on models that do not reason

Requests that set reasoning_effort on a model without reasoning (the gpt-4o and gpt-4.1 families) used to return the provider 400. The gateway now drops the field there, and maps a value a reasoning model does not accept (none on gpt-5-mini or o4-mini) to the nearest one it does. The adjustment is named in the X-Dropped-Params / X-Mapped-Params response headers (on a streaming response, for the primary backing of the model), so agent frameworks that always send the field (Hermes Agent among them) work on these models unchanged.

betaFeature

Contact & sovereign waitlist forms

Marketing CTAs (Contact us / Join sovereign waitlist) now open a structured form. Submissions are saved and acknowledged by our team.

betaFix

Docs accuracy pass

Replaced placeholder model IDs in the docs with IDs from the live catalog. Removed the JAIS page from public docs. Production base URL is no longer rewritten to localhost in code samples.

betaFeature

Open weights vs Proprietary metadata

Provider and per-backing weights-type metadata is now resolved and surfaced in the public catalog, dashboard model list, and model detail page. The signal answers "can these weights be self-hosted?" — not a software-license claim.

betaImprovement

Catalog filtering refresh

Replaced the confusing "minimum context window" slider with discrete context buckets (8K+ / 32K+ / 128K+). Public marketing pages now force dark mode for a consistent first paint.

betaFeature

Playground: conversation history

Added a "last N messages" history control to the playground so multi-turn context is actually sent with each request.

betaFeature

Provider control plane

Operator-facing provider catalog with per-provider sync attempts, credential rotation, smoke tests, and EGP-sell / upstream-USD-cost attribution on usage events.

betaFeature

Instant API key revocation

Revoke an API key from the dashboard and it stops working on the next request. Our team can also disable a user or suspend an organisation immediately when an account has to be stopped.

Changelog | SovereignEG