37 tok/s from Qwen3-Next-80B (48 GB) on a 32 GB laptop, 24 without drafts vs 2.8 for llama.cpp; reading 62 vs 16 tok/s. Faster than llama.cpp on all 13 models compared in Q4_K_M. LLMs bigger than your RAM, pure C, CPU or GPU.
C avx2ccpu-inferenceggufinference-enginelinuxllmlocal-llmmixture-of-expertsmoenvmequantizationqwen3simdspeculative-decodingvulkan
/* SPDX-License-Identifier: GPL-3.0-or-later
Copyright (C) 2026 Maurizio Cammalleri */
/*
* multipart.h - a multipart/form-data body (an upload) taken apart
* (multipart.c).
*/
#ifndef JANAS_SERVER_MULTIPART_H
#define JANAS_SERVER_MULTIPART_H
#include <stddef.h>
struct srv_part {
char name[64]; /* the form field */
char filename[256]; /* as the client sent it: never a path here */
int has_filename;
char type[128]; /* its Content-Type, or "" */
const char *data; /* into the body */
size_t n;
};
/* The parts of body (ctype: the request's Content-Type, its boundary in
it), at most max: how many, or -1 with the reason in err. */
int srv_multipart_parse(const char *ctype, const char *body, size_t n,
struct srv_part *parts, int max, char *err,
size_t err_len);
/* The part of that name, or NULL. */
const struct srv_part *srv_multipart_get(const struct srv_part *parts, int n,
const char *name);
#endif