From 1a3faa3ec450cfb601e8ef89a3d476994312195c Mon Sep 17 00:00:00 2001 From: alikarimirad Date: Tue, 22 Sep 2026 01:16:16 +0330 Subject: [PATCH] fix(opencode): add effort reasoning_options for mimo-v2.6 models MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Replace reasoning_options = [] with effort [none, low, medium, high] on opencode-go/mimo-v2.6-flash, opencode-go/mimo-v2.6-pro and opencode/mimo-v2.6-flash-free so clients show an effort picker. Evidence: live probe 2026-09-22 on POST /zen/go/v1/chat/completions — none|low|medium|high pass (none -> 0 reasoning tokens), while xhigh|max|minimal|adaptive|default -> 400. Free-tier values mirror the zen/go probe on the same base_model plus kilo's xiaomi/mimo-v2.6 entry. bun validate passes. --- providers/opencode-go/models/mimo-v2.6-flash.toml | 6 +++++- providers/opencode-go/models/mimo-v2.6-pro.toml | 6 +++++- providers/opencode/models/mimo-v2.6-flash-free.toml | 7 ++++++- 3 files changed, 16 insertions(+), 3 deletions(-) diff --git a/providers/opencode-go/models/mimo-v2.6-flash.toml b/providers/opencode-go/models/mimo-v2.6-flash.toml index 06b0dcf1faa..7000af41713 100644 --- a/providers/opencode-go/models/mimo-v2.6-flash.toml +++ b/providers/opencode-go/models/mimo-v2.6-flash.toml @@ -1,5 +1,9 @@ +# Effort: reasoning_effort = none|low|medium|high (live probe 2026-09-22: these four pass; +# xhigh|max|minimal|adaptive|default → 400; none → 0 reasoning tokens; thinking.type also +# honored, but none ∈ effort so no separate toggle) +# https://mimo.mi.com/docs/api/chat/openai-api (accessed 2026-09-22) base_model = "xiaomi/mimo-v2.6-flash" -reasoning_options = [] +reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }] [interleaved] field = "reasoning_content" diff --git a/providers/opencode-go/models/mimo-v2.6-pro.toml b/providers/opencode-go/models/mimo-v2.6-pro.toml index 7513e538ec8..a4768bd3cec 100644 --- a/providers/opencode-go/models/mimo-v2.6-pro.toml +++ b/providers/opencode-go/models/mimo-v2.6-pro.toml @@ -1,5 +1,9 @@ +# Effort: reasoning_effort = none|low|medium|high (live probe 2026-09-22: these four pass; +# xhigh|max|minimal|adaptive|default → 400; none → 0 reasoning tokens; thinking.type also +# honored, but none ∈ effort so no separate toggle) +# https://mimo.mi.com/docs/api/chat/openai-api (accessed 2026-09-22) base_model = "xiaomi/mimo-v2.6-pro" -reasoning_options = [] +reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }] [interleaved] field = "reasoning_content" diff --git a/providers/opencode/models/mimo-v2.6-flash-free.toml b/providers/opencode/models/mimo-v2.6-flash-free.toml index 753e0075aa2..ecdab2cb6d3 100644 --- a/providers/opencode/models/mimo-v2.6-flash-free.toml +++ b/providers/opencode/models/mimo-v2.6-flash-free.toml @@ -1,6 +1,11 @@ +# Effort: reasoning_effort = none|low|medium|high (Zen free tier rejects probes outside +# OpenCode — 403 FreeTierError — so values come from the live zen/go probe on the same +# base_model 2026-09-22 plus kilo's effort entry for xiaomi/mimo-v2.6-*; none ∈ effort so +# no separate toggle) +# https://mimo.mi.com/docs/api/chat/openai-api (accessed 2026-09-22) base_model = "xiaomi/mimo-v2.6-flash" name = "MiMo-V2.6-Flash Free" -reasoning_options = [] +reasoning_options = [{ type = "effort", values = ["none", "low", "medium", "high"] }] [interleaved] field = "reasoning_content"