Parameters by Upstream

Some request fields do not mean the same thing everywhere. This page states what the gateway does with each one on the way to each provider - passed through, translated into the provider's own equivalent, or dropped.

This is not a list of every accepted field; each endpoint page documents its own request body. Listed here are the fields where supported is not a yes or no, because the answer changes with the provider your model runs on.

`reasoning_effort`

Full guide, with values and cost impact →

You call	Upstream	Behaviour	Description
`/v1/chat/completions`	OpenAI-compatible (GPT, GLM, DeepSeek, Qwen, Kimi…)	transmis tel quel	Forwarded to the provider as sent, with one exception: when a model's catalog entry lists the effort levels it accepts, a value below the lowest is raised to it and one above the highest steps down (GPT-6.1 Sol cannot turn reasoning off, so none and minimal run at low). Otherwise the gateway neither validates the value nor guarantees the provider honours it. Commonly minimal, low, medium and high, but the accepted set is the provider's: OpenAI also takes none on most models, GLM takes up to max. Check the provider's own reference for its range.
`/v1/chat/completions`	Google Vertex - Gemini 4	converti	Becomes Gemini's `thinkingLevel`, kept inside the three levels Gemini 4 Argon accepts: none and minimal run at low (the model cannot turn thinking off), low/medium/high map straight across, xhigh and max run at high. An explicit `google.thinking_config` in the request wins over this mapping.
`/v1/chat/completions`	Google Vertex - Gemini 3.x	converti	Becomes Gemini's `thinkingLevel`: none and minimal both map to minimal, low/medium/high map straight across. An explicit `google.thinking_config` in the request wins over this mapping.
`/v1/chat/completions`	Google Vertex - Gemini 2.5	ignoré	Gemini 2.5 only understands `thinkingBudget` and rejects `thinkingLevel` outright, so the gateway sends neither and leaves the model at its default thinking behaviour. Use `google.thinking_config` with `thinking_budget` to control 2.5.
`/v1/chat/completions`	Anthropic (Claude)	converti	Devient le `thinking` d'Anthropic, sous la forme que le modèle accepte. Sur les modèles Claude qui n'acceptent que la réflexion adaptative (Claude Opus 4.7 et suivants, Sonnet 5 et suivants, Haiku 5.5), low/medium/high/xhigh/max deviennent `thinking:{"type":"adaptive"}` avec `output_config.effort` au même niveau, et minimal s'exécute en low. none désactive la réflexion sous la forme propre au modèle : `{"type":"disabled"}` sur la plupart, `{"type":"between_tools"}` sur Sonnet 5.5, et adaptatif en low sur Opus 5.5, qui ne peut pas désactiver la réflexion. Toute autre valeur n'est pas transmise. Sur les modèles Claude plus anciens, elle devient un budget : none et minimal désactivent la réflexion (`{"type":"disabled"}`) ; low/medium/high/xhigh/max deviennent un budget de 2,048 / 8,192 / 16,384 / 32,768 tokens, l'inverse du découpage utilisé dans l'autre sens. Un budget dérivé est plafonné sous `max_tokens`, et n'est pas envoyé du tout quand `max_tokens` ne peut pas contenir le minimum de 1,024 tokens d'Anthropic : demander un niveau d'effort ne transforme donc jamais une requête qui fonctionne en erreur 400 de l'amont.
`/v1/messages`	OpenAI-compatible (GPT, GLM, DeepSeek, Qwen, Kimi…)	converti	Derived from `thinking.budget_tokens` by bucket, when extended thinking is enabled and the budget is above zero: up to 2,048 tokens becomes low, up to 8,192 becomes medium, above that becomes high. A zero or absent budget sends nothing. For a model whose catalog entry lists the effort levels it accepts (GPT-6.1 Sol), the request's intent is followed instead: thinking disabled or between_tools runs at the lowest level (low), even when output_config.effort is also set; otherwise output_config.effort is sent as the effort.
`/v1/responses`	OpenAI-compatible (GPT, GLM, DeepSeek, Qwen, Kimi…)	converti	Read out of the Responses-API `reasoning.effort` object and sent as the flat `reasoning_effort` field.

`thinking / thinking_budget / enable_thinking`

You call	Upstream	Behaviour	Description
`/v1/chat/completions`	Anthropic (Claude)	converti	Ce sont les noms que d'autres fournisseurs donnent au même réglage, et la passerelle les normalise. `thinking:{"type":"disabled"}`, `enable_thinking:false` et `thinking_budget:0` désactivent tous la réflexion, comme reasoning_effort none sur ce modèle. Sur les modèles Claude plus anciens, `thinking_budget:N` et `thinking:{"type":"enabled","budget_tokens":N}` transmettent le budget tel quel ; une valeur hors limites est refusée par l'amont plutôt qu'ajustée en silence. Sur les modèles qui n'acceptent que la réflexion adaptative, aucun budget n'est envoyé : il devient `output_config.effort` (jusqu'à 2,048 low, jusqu'à 8,192 medium, sous 32,768 high, 32,768 et plus xhigh ; jamais max). `thinking:{"type":"adaptive"}` seul n'envoie aucun réglage de réflexion, et le modèle s'exécute avec son comportement par défaut.
`/v1/chat/completions`	Google Vertex - Gemini 4	converti	Normalised into `thinkingConfig.thinkingLevel`, because Gemini 4 Argon rejects `thinkingBudget`: turning thinking off runs at low, and an explicit budget is folded into a level (up to 2,048 tokens low, up to 8,192 medium, above that high). An explicit `google.thinking_config` in the request still wins.
`/v1/chat/completions`	Google Vertex - Gemini 3.x	converti	Normalised into `thinkingConfig`: turning thinking off becomes `thinkingLevel:"minimal"`, and an explicit budget becomes `thinkingBudget`. An explicit `google.thinking_config` in the request still wins over all of it.
`/v1/chat/completions`	Google Vertex - Gemini 2.5	converti	Normalised into `thinkingConfig.thinkingBudget` - turning thinking off sends a budget of 0, and an explicit budget passes through. `thinkingLevel` is not used here because 2.5 rejects it. An explicit `google.thinking_config` in the request still wins.

Every row is transcribed from the converter that implements it and pinned by a Go test, so this table cannot drift from the running gateway without a build failing.