libjinfer and the jinfer command
Copyright 2026 Quixotic AI (https://qxotic.ai)

Their own code is licensed under the Apache License, Version 2.0 (see LICENSE).

Oracle GraalVM (https://www.graalvm.org)
  The binaries in the release archives and in the jinfer wheel are native images built with
  Oracle GraalVM. They contain Oracle's Native Image runtime and the parts of the Java class
  library they use, redistributed under the GraalVM Free Terms and Conditions (GFTC),
  https://www.oracle.com/downloads/licenses/graal-free-license.html, with Oracle's notices in
  them intact. The GFTC counts those portions of a native image as the unmodified
  Program and permits redistributing it "provided that You do not charge Your licensees any
  fees associated with such distribution or use of the Program, including, without
  limitation, fees for products that include or are bundled with a copy of the Program".
  What that means: you may use these binaries as they are, in commercial products and paid
  services included. Charging a fee for a product that bundles a copy is not within the GFTC;
  that takes a license from Oracle, or a build with GraalVM CE, see README.md, which carries
  the runtime under the GNU General Public License version 2 with the Classpath Exception and
  no such condition.

llama.cpp and ggml (https://github.com/ggml-org/llama.cpp, https://github.com/ggml-org/ggml)
  Copyright (c) 2023-2024 The ggml authors. MIT License.
  - The pretokenizer regular expressions and their family groups mirror llama-vocab.cpp.
  - The Q1_0 AVX2 dot product is a port of ggml_vec_dot_q1_0_q8_0, the NEON Q1_0 kernel reuses
    ggml's table_b2b_0 sign-expansion construction, and the K-quant block layouts and decode
    arithmetic follow ggml's quants.
  - The Qwen 3.5 vision tower is a port of clip_graph_qwen3vl.
  - The Gemma 4 multi-token-prediction head follows the structure of gemma4-assistant.cpp.

  Permission is hereby granted, free of charge, to any person obtaining a copy of this
  software and associated documentation files (the "Software"), to deal in the Software
  without restriction, including without limitation the rights to use, copy, modify, merge,
  publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons
  to whom the Software is furnished to do so, subject to the following conditions: The above
  copyright notice and this permission notice shall be included in all copies or substantial
  portions of the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
  EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
  FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
  HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
  CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
  USE OR OTHER DEALINGS IN THE SOFTWARE.

llamafile tinyBLAS (https://github.com/Mozilla-Ocho/llamafile)
  Copyright 2024 Mozilla Foundation. Apache License, Version 2.0.
  - The register-tiled dense GEMM kernels and their recursive tile selection (mnpack) follow
    tinyBLAS's design.

External programs (not distributed with this product)
  - espeak-ng (GPL-3.0) or espeak: the speech ports invoke a binary found on the PATH as a
    phonemizer. It is neither bundled nor required.
