From d9b7ca9cc48889d0844c8d11dd0781eb700e14c4 Mon Sep 17 00:00:00 2001 From: Lukasz Raczylo Date: Sun, 30 Aug 2026 14:19:22 +0100 Subject: [PATCH] feat(config): raise max_tokens_default to 40000 At 25000 the heaviest thinking-model tests exhausted the whole budget on reasoning and scored zero with an empty answer (8 truncated cells in the 2026-08-30 3-model run, e.g. ts-eslint-flat-config on qwen3.8-flash-next at exactly 25000 completion tokens). --- internal/config/config.go | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/internal/config/config.go b/internal/config/config.go index f05b05f..9f82339 100644 --- a/internal/config/config.go +++ b/internal/config/config.go @@ -15,7 +15,10 @@ import ( const apiKeyEnvVar = "LLMTB_API_KEY" // #nosec G101 -- this is an env var *name*, not a credential value // Defaults applied when the corresponding YAML key is omitted (or explicit -// zero). max_tokens_default was raised from 4000 to 12000 after a live run +// zero). max_tokens_default was raised from 4000 to 12000, then to 40000 +// (2026-08-30: at 25000 the heaviest thinking-model tests still exhausted +// the whole budget on reasoning and scored 0 with an empty answer, e.g. +// ts-eslint-flat-config on qwen3.8-flash-next), after a live run // showed reasoning models spend thousands of completion tokens on // reasoning_content before writing an answer to content: a small // per-test/default budget truncates the answer (finish_reason=length) @@ -23,7 +26,7 @@ const apiKeyEnvVar = "LLMTB_API_KEY" // #nosec G101 -- this is an env var *name* const ( defaultConcurrency = 8 defaultRequestTimeout = 300 * time.Second - defaultMaxTokens = 12000 + defaultMaxTokens = 40000 ) // Config is the top-level configuration loaded from config.yaml.