From 32ff96e7282535913593375bcacd517ee0df88f8 Mon Sep 17 00:00:00 2001 From: Nils Fedrigo <185288491+Nils-Fedrigo@users.noreply.github.com> Date: Wed, 12 Aug 2026 12:13:06 +0200 Subject: [PATCH 1/2] docs: clarify where to find model tier breakdown --- .../usage-seats-and-credits/model-access-tiers.mdx | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-) diff --git a/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx b/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx index c32bb5f..c9535eb 100644 --- a/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx +++ b/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx @@ -23,6 +23,16 @@ Tiers are cumulative. Granting **Up to Balanced** also grants Cost efficient. Gr A tier applies to a model **and** a reasoning effort, not to a model alone. The same model can sit in two tiers: for example a mid-size model at Light effort can be Cost efficient, at Medium effort Balanced, and at High effort Premium. A frontier model is Premium at every effort. +## See the models in each tier + +To see the current model and reasoning-effort combinations included in each tier, go to **Admin** > **Usage** and click the **ⓘ** icon next to **Models tier**. The breakdown is available in all three places where model-tier access is displayed: + +- **Members** tab: next to the **Models tier** column heading. +- **Groups** tab: next to the **Models tier** column heading. +- **Settings** tab: next to the **Models tier** section heading. + +The list in the Usage page is the source of truth. It changes as Dust adds models and providers update pricing or capabilities. + ## Set the workspace ceiling 1. Go to **Admin** > **Usage**. @@ -82,7 +92,7 @@ No. Every workspace starts at **Up to Premium**, which restricts nothing. Config ### Which models are Premium, and which are Cost efficient? -It depends on the model and the reasoning effort. You can get the full list by clicking on the (i) icon near model tier selection, in the Usage page. As a rule of thumb: small and mini models are Cost efficient, mid-size models are Cost efficient or Balanced at low effort and Premium at their highest effort, and frontier models are Premium at every effort. The tiers evolve as providers release new models and change prices. +It depends on the model and the reasoning effort. See [the model-tier breakdown](#see-the-models-in-each-tier) in **Admin** > **Usage** for the current list. As a rule of thumb: small and mini models are Cost efficient, mid-size models are Cost efficient or Balanced at low effort and Premium at their highest effort, and frontier models are Premium at every effort. The tiers evolve as providers release new models and change prices. ### Does restricting tiers break agents that were already built on premium models? From cd2a5298c5c541c181fcaf65a79c8e02c8a27ec4 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?L=C3=A9andre=20Le=20Bizec?= <95234460+LeandreLeBizec@users.noreply.github.com> Date: Wed, 12 Aug 2026 15:28:01 +0200 Subject: [PATCH 2/2] docs: add current model tier breakdown --- .../model-access-tiers.mdx | 35 +++++++++++++++++++ 1 file changed, 35 insertions(+) diff --git a/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx b/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx index c9535eb..46c86f8 100644 --- a/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx +++ b/docs/user-documentation/admins/usage-seats-and-credits/model-access-tiers.mdx @@ -33,6 +33,41 @@ To see the current model and reasoning-effort combinations included in each tier The list in the Usage page is the source of truth. It changes as Dust adds models and providers update pricing or capabilities. +### Current tier breakdown + +The following table reflects the model catalog and tier assignments as of August 12, 2026. A model can appear in more than one tier because its tier depends on the reasoning effort. `All efforts` means all reasoning efforts supported by that model. + +| Model | Cost efficient | Balanced | Premium | +| --- | --- | --- | --- | +| GPT 5.6 Sol | — | `none` | `light`, `medium`, `high` | +| GPT 5.6 Terra | — | all efforts | — | +| GPT 5.6 Luna | `none`, `light` | `medium`, `high` | — | +| GPT-5.4 Mini | all efforts | — | — | +| GPT-5.4 Nano | all efforts | — | — | +| Claude 4.5 Haiku | all efforts | — | — | +| Claude Sonnet 5 | `light` | `medium` | `high` | +| Claude Sonnet 4.6 | `light` | `medium` | `high` | +| Claude Opus 5 | — | — | all efforts | +| Claude Opus 4.8 | — | — | all efforts | +| Claude Fable 5 | — | — | all efforts | +| Mistral Large | — | `none` | — | +| Mistral Medium 3.5 | — | `none` | `high` | +| Mistral Small | `none` | — | — | +| Mistral Codestral | `none` | — | — | +| Gemini 3.1 Flash Lite | `none`, `light`, `medium` | `high` | — | +| Gemini 3.1 Pro (Preview) | — | `light` | `medium`, `high` | +| Gemini 3.5 Flash Lite | `none`, `light`, `medium` | `high` | — | +| Gemini 3.5 Flash | — | `light`, `medium` | `high` | +| Gemini 3.6 Flash | — | `light`, `medium` | `high` | +| DeepSeek V4 Pro (Fireworks) | — | `none` | — | +| DeepSeek V4 Flash (Fireworks) | all efforts | — | — | +| Kimi K2.6 (Fireworks) | — | all efforts | — | +| Kimi K3 (Fireworks) | — | `light` | `medium`, `high` | +| GLM-5.2 (Fireworks) | — | `high` | — | +| Grok 4.5 | — | all efforts | — | + +The table covers the models currently shown in the Usage-page breakdown. Workspace plan, provider access, region, and other availability settings can further limit which models a member can select. + ## Set the workspace ceiling 1. Go to **Admin** > **Usage**.