v3.14.0

3.14.0 (2025-10-02)

Features

• Qwen3 Reranker support (#506) (00305f7) (see #506 for prequantized Qwen3 Reranker models you can use)

Bug Fixes

• handle HuggingFace rate limit responses (#506) (00305f7)
• adapt to llama.cpp breaking changes (#506) (00305f7)

---

Shipped with llama.cpp release b6673

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.13.0

3.13.0 (2025-09-09)

Features

• Seed OSS support (#502) (eefe78c)

Bug Fixes

• adapt to breaking llama.cpp changes (#501) (76b505e)
• Vulkan: read external memory usage (#500) (d33cc31)

---

Shipped with llama.cpp release b6431

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.12.4

✨ gpt-oss is here! ✨

Read about the release in the blog post

---

3.12.4 (2025-08-28)

Bug Fixes

• gpt-oss prompt preloading (#496) (db4a243)

---

Shipped with llama.cpp release b6301

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.12.3

✨ gpt-oss is here! ✨

Read about the release in the blog post

---

3.12.3 (2025-08-26)

Bug Fixes

• Vulkan: context creation edge cases (#492) (12749c0)
• prebuilt binaries CUDA 13 support (#494) (b10999d)
• don't share loaded shared libraries between backends (#492) (12749c0)
• split prebuilt CUDA binaries into 2 npm modules (#495) (6e59160)

---

Shipped with llama.cpp release b6294

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.12.1

✨ gpt-oss is here! ✨

Read about the release in the blog post

---

3.12.1 (2025-08-11)

Features

• comment segment budget (#489) (30eaa23) (documentation: API: LLamaChatPromptOptions["budgets"]["commentTokens"])
• Electron template: comment segments
• Electron template: improve completions speed when using functions

Bug Fixes

• gpt-oss segment budgets (#489) (30eaa23)
• add support for more gpt-oss variations (#489) (30eaa23)
• default to using a model message for prompt completion on unsupported models (#489) (30eaa23)
• prompt completion config (#490) (f849cd9)

---

Shipped with llama.cpp release b6133

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.12.0

✨ gpt-oss is here! ✨

Read about the release in the blog post

---

3.12.0 (2025-08-09)

Features

• gpt-oss support (#487) (722e29d) (documentation: gpt-oss)

Bug Fixes

• Llama: expose the numa (#485) (ea0d815)
• add --numa flag to cli commands (#485) (ea0d815)

---

Shipped with llama.cpp release b6122

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.11.0

3.11.0 (2025-07-29)

Features

• NUMA policy (#482) (a2ddaa2) (documentation: API: LlamaOptions["numa"])
• inspect gpu command: log prebuilt binaries and cloned source releases (#482) (a2ddaa2)

Bug Fixes

• add missing GGUF metadata types (#482) (a2ddaa2)
• level of some internal logs (#482) (a2ddaa2)
• JSON schema grammar edge case (#482) (a2ddaa2)

---

Shipped with llama.cpp release b6018

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.10.0

3.10.0 (2025-06-12)

Features

• JSON Schema Grammar: $defs and $ref support with full inferred types (#472) (9cdbce9)
• inspect gguf command: format and print the Jinja chat template with --key .chatTemplate (#472) (9cdbce9)

Bug Fixes

• JinjaTemplateChatWrapper: first function call prefix detection (#472) (9cdbce9)
• QwenChatWrapper: improve Qwen chat template detection (#472) (9cdbce9)
• apply maxTokens on function calling parameters (#472) (9cdbce9)
• adjust default prompt completion length based on SWA size when relevant (#472) (9cdbce9)
• improve thought segmentation syntax extraction (#472) (9cdbce9)
• adapt to llama.cpp changes (#472) (9cdbce9)

---

Shipped with llama.cpp release b5640

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.9.0

3.9.0 (2025-06-04)

Features

• reasoning budget (#468) (ea8d904) (documentation: Set Reasoning Budget)
• SWA (Sliding Window Attention) support - greatly reduced context memory consumption on supported models (#468) (ea8d904)
• documentation: LLMs friendly llms.md and llms-full.md files (#468) (ea8d904)

Bug Fixes

• prompt completion edge cases (#468) (ea8d904)
• adapt to llama.cpp changes (#468) (ea8d904)

---

Shipped with llama.cpp release b5590

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)

v3.8.1

3.8.1 (2025-05-19)

Bug Fixes

• getLlamaGpuTypes: edge case (#463) (1799127)
• remove prompt completion from the cached context window (#463) (1799127)

---

Shipped with llama.cpp release b5415

To use the latest llama.cpp release available, run npx -n node-llama-cpp source download --release latest. (learn more)