Fix merge conflicts
keypair34 committed
Apr 26, 2026 at 09:16 UTC
53b8844e4bc54a7802f80b5f218b139b3929a0e3
2 files changed
+16
-16
src/inference/engine.rs
+14
-16
@@ -2003,8 +2003,21 @@ impl GgufModelConfig {
2003
}
2004
}
2005
2006
- /// Qwen 3 8B Instruct (GGUF Q4_K_M) — ~5 GB.
2006
+ /// Qwen 3 14B Instruct (GGUF Q4_K_M) — ~8.4 GB.
2007
///
2008
+ /// Strongest reasoning and tool-calling model with extended thinking.
2009
+ /// Best all-around model for macOS with 16+ GB RAM.
2010
+ pub fn qwen3_14b() -> Self {
2011
+ Self {
2012
+ model_id: super::models::BARTOWSKI_QWEN3_14B_GGUF.into(),
2013
+ files: vec![super::models::QWEN3_14B_GGUF_FILE.into()],
2014
+ tok_model_id: None,
2015
+ display_name: "Qwen 3 14B (Q4_K_M)".into(),
2016
+ approx_memory: "~8.4 GB".into(),
2017
+ chat_template: None,
2018
+ }
2019
+ }
2020
+
2021
/// Strong tool-calling model with extended thinking. Best balance of
2022
/// quality and memory for macOS with 24+ GB RAM.
2023
pub fn qwen3_8b() -> Self {
@@ -2018,21 +2031,6 @@ impl GgufModelConfig {
2031
}
2032
}
2033
2021
- /// Qwen 3 14B (GGUF Q4_K_M) — ~8.4 GB.
2022
- ///
2023
- /// Strong reasoning and tool-calling model with extended thinking.
2024
- /// Best all-around model for macOS with 16+ GB RAM.
2025
- pub fn qwen3_14b() -> Self {
2026
- Self {
2027
- model_id: super::models::BARTOWSKI_QWEN3_14B_GGUF.into(),
2028
- files: vec![super::models::QWEN3_14B_GGUF_FILE.into()],
2029
- tok_model_id: None,
2030
- display_name: "Qwen 3 14B (Q4_K_M)".into(),
2031
- approx_memory: "~8.4 GB".into(),
2032
- chat_template: None,
2033
- }
2034
- }
2035
-
2034
/// DeepSeek Coder v1 6.7B Instruct (GGUF Q4_K_M) — ~3.8 GB.
2035
///
2036
/// Strong code generation model using the `llama` GGUF architecture.
src/inference/models.rs
+2
@@ -204,6 +204,8 @@ pub const SUPPORTED_MODEL_INFO: &[SupportedModelInfo] = &[
204
org: "Qwen / Alibaba",
205
description: "Strong reasoning and tool-calling model with extended thinking (~8.4 GB). \
206
Best all-around model for macOS with 16+ GB RAM.",
207
+ // Qwen_Qwen3-14B-Q4_K_M.gguf from bartowski repo.
208
+ // Exact file size from HuggingFace API siblings[].size.
209
expected_size_bytes: 9_001_753_632,
210
},
211
SupportedModelInfo {