From d2b5674e186cc1ab93f1e98802d424363ff058d3 Mon Sep 17 00:00:00 2001 From: Antonio-Faure <209759495+Antonio-Faure@users.noreply.github.com> Date: Sat, 31 Oct 2026 20:26:35 +0100 Subject: [PATCH 1/4] docs(console): describe Go usage as a shared allowance The Go page said each model had its own 5-hour, weekly and monthly limits. The service enforces one workspace-wide allowance across all paid models, with each model's monthly limit weighting how heavily its usage counts. --- services/site/src/docs/content/console/go.mdx | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/services/site/src/docs/content/console/go.mdx b/services/site/src/docs/content/console/go.mdx index d6e6ed54f593..be37e3a70e5f 100644 --- a/services/site/src/docs/content/console/go.mdx +++ b/services/site/src/docs/content/console/go.mdx @@ -136,11 +136,11 @@ Usage limits are defined as monthly dollar amounts. The table below shows the monthly limit for each plan and the token costs for each model. Token pricing is the same for Go, Go Plus, and Go Teams. -Each model has the following usage limits: 5-hour — 20% of the monthly limit; -weekly — 50%; and monthly — 100%. +All paid models share one workspace-wide allowance: 5-hour — 20% of the monthly +allowance; weekly — 50%; and monthly — 100%. -Each model's monthly limit below determines how its usage counts toward those allowances. Each response reports how -much the request counted, as [`opencode.allowance`](/console/api/inference#cost). +Each model's monthly limit below determines how heavily its usage counts toward that shared allowance. Each response +reports how much the request counted, as [`opencode.allowance`](/console/api/inference#cost). Token prices are per 1M tokens. From 46292867ebcccefff83fc5a895d0d266c341dc1a Mon Sep 17 00:00:00 2001 From: Antonio-Faure <209759495+Antonio-Faure@users.noreply.github.com> Date: Sat, 31 Oct 2026 20:28:39 +0100 Subject: [PATCH 2/4] docs(console): address shared allowance review Clarify that the allowance is shared across models but not across Go Teams members, note that the estimates assume one model, and spell out the weighting and the block behaviour. --- services/site/src/docs/content/console/go.mdx | 19 +++++++++++++------ 1 file changed, 13 insertions(+), 6 deletions(-) diff --git a/services/site/src/docs/content/console/go.mdx b/services/site/src/docs/content/console/go.mdx index be37e3a70e5f..8779c96736b9 100644 --- a/services/site/src/docs/content/console/go.mdx +++ b/services/site/src/docs/content/console/go.mdx @@ -136,11 +136,17 @@ Usage limits are defined as monthly dollar amounts. The table below shows the monthly limit for each plan and the token costs for each model. Token pricing is the same for Go, Go Plus, and Go Teams. -All paid models share one workspace-wide allowance: 5-hour — 20% of the monthly -allowance; weekly — 50%; and monthly — 100%. +All paid models share one allowance, weighted by each model's monthly limit: the +5-hour window allows 20% of the monthly allowance, the weekly window 50%, and the +monthly window 100%. On Go Teams, each member has their own allowance; it is not +shared across members. -Each model's monthly limit below determines how heavily its usage counts toward that shared allowance. Each response -reports how much the request counted, as [`opencode.allowance`](/console/api/inference#cost). +Each model's monthly limit below determines how heavily its usage counts toward that shared allowance. For example, +on Go, $1 on a model with a $15 monthly limit counts as $4 of the $60 allowance. Each response reports how much the +request counted, as [`opencode.allowance`](/console/api/inference#cost). + +When any of the three limits is reached, every paid model is blocked until that limit resets. Free models continue to +work. Token prices are per 1M tokens. @@ -297,8 +303,9 @@ Token prices are per 1M tokens. ### Estimated requests -The tables below estimate request counts based on typical Go usage patterns. Go -Plus and Go Teams estimates scale with each model's higher usage limit. +The tables below estimate request counts based on typical Go usage patterns, +assuming you use only that model. Go Plus and Go Teams estimates scale with each +model's higher usage limit.
From 9b1f40200bc6c58b977825e01db6232956313f51 Mon Sep 17 00:00:00 2001 From: Antonio-Faure <209759495+Antonio-Faure@users.noreply.github.com> Date: Sat, 31 Oct 2026 20:29:52 +0100 Subject: [PATCH 3/4] docs(console): call the monthly limit a weight Make explicit that the Monthly limit column is a weighting for the shared allowance, not a separate per-model budget. --- services/site/src/docs/content/console/go.mdx | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/services/site/src/docs/content/console/go.mdx b/services/site/src/docs/content/console/go.mdx index 8779c96736b9..2c2548674ce1 100644 --- a/services/site/src/docs/content/console/go.mdx +++ b/services/site/src/docs/content/console/go.mdx @@ -141,9 +141,11 @@ All paid models share one allowance, weighted by each model's monthly limit: the monthly window 100%. On Go Teams, each member has their own allowance; it is not shared across members. -Each model's monthly limit below determines how heavily its usage counts toward that shared allowance. For example, -on Go, $1 on a model with a $15 monthly limit counts as $4 of the $60 allowance. Each response reports how much the -request counted, as [`opencode.allowance`](/console/api/inference#cost). +Each model's monthly limit below determines how heavily its usage counts toward +that shared allowance; the **Monthly limit** column is a weight, not a separate +budget. For example, on Go, $1 on a model with a $15 monthly limit counts as $4 of +the $60 allowance. Each response reports how much the request counted, as +[`opencode.allowance`](/console/api/inference#cost). When any of the three limits is reached, every paid model is blocked until that limit resets. Free models continue to work. From a9d8841cbe7727ed8d754c891d227fea85c20bf3 Mon Sep 17 00:00:00 2001 From: vprdev Date: Sun, 15 Nov 2026 19:45:12 -0800 Subject: [PATCH 4/4] docs(go): explain model weights and automatic title usage --- services/site/src/docs/content/console/go.mdx | 30 +++++++++---------- 1 file changed, 14 insertions(+), 16 deletions(-) diff --git a/services/site/src/docs/content/console/go.mdx b/services/site/src/docs/content/console/go.mdx index 2c2548674ce1..9969bb01396e 100644 --- a/services/site/src/docs/content/console/go.mdx +++ b/services/site/src/docs/content/console/go.mdx @@ -132,25 +132,23 @@ investigated. The linked reports track fixes and workarounds. ## Usage limits -Usage limits are defined as monthly dollar amounts. The table below shows the -monthly limit for each plan and the token costs for each model. Token pricing is -the same for Go, Go Plus, and Go Teams. - -All paid models share one allowance, weighted by each model's monthly limit: the -5-hour window allows 20% of the monthly allowance, the weekly window 50%, and the -monthly window 100%. On Go Teams, each member has their own allowance; it is not -shared across members. - -Each model's monthly limit below determines how heavily its usage counts toward -that shared allowance; the **Monthly limit** column is a weight, not a separate -budget. For example, on Go, $1 on a model with a $15 monthly limit counts as $4 of -the $60 allowance. Each response reports how much the request counted, as +All priced Go models share one allowance. You can use 20% of a month's allowance in the 5-hour window, 50% in the +weekly window and 100% in the monthly window. On Go Teams, each member has their own allowance across models. + +**Monthly limit** is the dollar amount of usage that would use your whole monthly allowance if you used only that +model. For example, $1 of usage on a model with a $15 monthly limit uses about 6.67% of the month. That's the same +share as about $6.67 on a model with a $100 monthly limit. Each response reports its allowance usage in [`opencode.allowance`](/console/api/inference#cost). -When any of the three limits is reached, every paid model is blocked until that limit resets. Free models continue to -work. +OpenCode makes a separate request to title a session. When it uses Go, that request counts at the title model's rate. +OpenCode can select GPT-6 Luna when it's available on Go, even if you picked another model for the conversation. +In OpenCode V2, choose the title model with `agents.title.model` in your [agent configuration](/agents). +If you're [migrating a V1 configuration](/migrate-v1), this replaces `small_model`. + +Switching priced models leaves all three meters where they are. If any window is full, included usage resumes once +every full window has reset. You can still use free models or enable [Extra Usage](#usage-beyond-limits). -Token prices are per 1M tokens. +Token prices below are per 1M tokens and are the same for Go, Go Plus and Go Teams.