alex/g3
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix embedded provider max_tokens default (2048 -> 8192)

Browse Source 
The resolve_max_tokens() function was returning 2048 for embedded providers,
which caused responses to be truncated prematurely. Increased to 8192 to
allow the provider's own effective_max_tokens() calculation to work properly.
This commit is contained in:
Dhanji R. Prasanna
2026-01-28 13:58:14 +11:00
parent 58fe74334d
commit fe33568ee0
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
crates/g3-core/src/provider_config.rs
View File

@@ -72,7 +72,7 @@ pub fn resolve_max_tokens(config: &Config, provider_name: &str) -> u32 {
"anthropic" => 32000, // Anthropic provider defaults to 32768, we use 32000 "anthropic" => 32000, // Anthropic provider defaults to 32768, we use 32000
"databricks" => 32000, // Databricks is passthru to Anthropic, match its defaults "databricks" => 32000, // Databricks is passthru to Anthropic, match its defaults
"openai" => 32000, // OpenAI models support large outputs "openai" => 32000, // OpenAI models support large outputs
"embedded" => 2048, // Embedded provider defaults to 2048 "embedded" => 8192, // Embedded provider: let provider's effective_max_tokens() handle it
_ => 16000, // Generic fallback _ => 16000, // Generic fallback
}; };
let base = get_max_tokens(config, provider_name).unwrap_or(provider_default); let base = get_max_tokens(config, provider_name).unwrap_or(provider_default);