Vulnerabilities
All
My stack
Known exploited
Public exploit
Critical (9+)
Likely exploited (EPSS ≥ 50%)
Changed, 24 h
New, 24 h
Sort:
risk
modified
published
CVSS
EPSS
| CVE | CVSS | EPSS | Vendor, product | Description | Modified |
|---|---|---|---|---|---|
| CVE-2026-34159exploit ×3 | 9.8 critical | 2.0% | ggml, llama.cpp | llama.cpp is an inference of several LLM models in C/C++. Prior to version b8492, the RPC backend's deserialize_tensor() skips all bounds validation when a tensor's buffer field is 0. An unauthenticated attacker can read and write arbitrary process memory via crafted GRAPH_COMPUTE messages. Combined with pointer leaks from ALLOC_BUFFER/BUFFER_GET_BASE, this gives full ASLR bypass and remote code execution. No authentication required, just TCP access to the RPC server port. This issue has been patched in version b8492. | Mon 26 Oct |
| CVE-2026-70638exploit | 8.5 high | 0.00% | ggml, llama.cpp | llama.cpp builds b1886 through b7445 contain an integer overflow vulnerability in the LLaMA-Android JNI wrapper where the new_1batch() function multiplies sizeof(llama_seq_id) by an attacker-controlled n_seq_max parameter without overflow validation, causing heap buffer allocation to wrap and allocate insufficient memory. Attackers can exploit this by providing a crafted n_seq_max value through a malicious model file or JNI call to trigger heap corruption and achieve denial of service or arbitrary code execution on Android applications using the LLaMA-Android binding. | Sat 24 Oct |
| CVE-2026-43629 | 9.2 critical | 1.0% | ggml, llama.cpp | llama.cpp builds b4882 through b9058 contain a heap buffer overflow vulnerability in the KV cache state restore path where the state_read_data() function computes write size without overflow checking, allowing attackers with write access to the slot_save_path directory to corrupt heap memory. Attackers can craft malicious state files where cell_count multiplication overflows or exceeds tensor buffer allocation to write attacker-controlled bytes past buffer boundaries, potentially resulting in heap metadata corruption, model weight corruption, or arbitrary code execution via function pointer overwrite. | Fri 23 Oct |
| CVE-2026-43631 | 9.2 critical | 1.0% | ggml, llama.cpp | llama.cpp builds b7492 through the latest b9060 contains a use-after-free vulnerability in the vocab pointer of llama-server when the --sleep-idle-seconds feature is enabled, allowing unauthenticated remote attackers to execute arbitrary code. Attackers can trigger the vulnerability by sending requests to affected endpoints while the server transitions to sleep mode, causing concurrent worker threads to dereference a freed vocab pointer that can be reclaimed with attacker-controlled data to achieve remote code execution. | Fri 23 Oct |
| CVE-2026-116467 | 6.9 medium | 1.0% | no CPE yet | A security flaw has been discovered in ggml-org llama.cpp up to 0.7.0. Affected by this vulnerability is an unknown functionality of the file src/llama-sampler.cpp of the component Mirostat Sampler. The manipulation results in reachable assertion. It is possible to launch the attack remotely. The project was informed of the problem early through a pull request but has not reacted yet. | Tue 27 Oct |
| CVE-2026-17500 | 6.9 medium | 1.0% | no CPE yet | A vulnerability was detected in ggml-org llama.cpp d006858/e15efe0. This affects the function _visit_pattern of the file common/json-schema-to-grammar.cpp. The manipulation results in null pointer dereference. The attack can be launched remotely. The pull request to fix this issue awaits acceptance. | Sat 24 Oct |
| CVE-2026-17501 | 6.9 medium | 1.0% | no CPE yet | A flaw has been found in ggml-org llama.cpp e15efe0. This vulnerability affects the function transform of the file common/json-schema-to-grammar.cpp of the component JSON-Schema-to-GBNF Conversion. This manipulation causes uncontrolled recursion. The attack may be initiated remotely. The pull request to fix this issue awaits acceptance. | Sat 24 Oct |
| CVE-2026-78147 | 6.9 medium | 1.0% | no CPE yet | A vulnerability was found in ggml-org llama.cpp bec4772f6. The impacted element is the function deserialize_tensor of the file ggml/src/ggml-rpc/ggml-rpc.cpp of the component ggml-RPC Server. Performing a manipulation of the argument op/op_params results in deserialization. The attack may be initiated remotely. This vulnerability is distinct from CVE-2026-34159 (GHSA-j8rj-fmpv-wcxw, PR #20908), which only added a buffer==nullptr rejection in create_node() and does not validate op or op_params. The reported GitHub issue was closed automatically due to inactivity. | 12 Oct |
| CVE-2026-78148 | 6.9 medium | 1.0% | no CPE yet | A vulnerability was determined in ggml-org llama.cpp bec4772f6. This affects the function rpc_server::graph_compute of the file ggml/src/ggml-rpc/ggml-rpc.cpp of the component ggml-RPC Server. Executing a manipulation can lead to null pointer dereference. The attack may be launched remotely. The pull request to fix this issue awaits acceptance. | 12 Oct |
| CVE-2026-118375 | 9.2 critical | — | no CPE yet | llama.cpp b8841 before b11865 contains an out-of-bounds write vulnerability in the ggml-rpc RDMA transport rdma_recv() function that allows unauthenticated peers to overflow the rpc-server stack. Attackers advertising RDMA capabilities in the HELLO message can send up to 256 KiB into a 1-byte stack variable in rpc_serve_client(), crashing the server or potentially executing code. | 20 h ago |
| CVE-2026-43632 | 9.2 critical | 0.00% | ggml, llama.cpp | llama.cpp builds b7492 through the latest b9060 contains a use-after-free vulnerability in llama-server affecting six tokenization endpoints (/tokenize, /detokenize, /infill, /apply-template, /rerank, and /anthropic/count_tokens) that bypass the task queue and access ctx_server.vocab directly on HTTP worker threads. Attackers can exploit a time-of-check-time-of-use race condition where the main thread destroys and frees vocab after the synchronization lock is released but before the handler finishes using it, causing a crash or potential code execution when --sleep-idle-seconds is configured. | Fri 23 Oct |
| CVE-2024-58400 | 8.8 high | 0.00% | no CPE yet | llama.cpp before b3701 contains an out-of-bounds read vulnerability in llama_decode_internal and llama_encode_internal that allows remote unauthenticated attackers to read memory by supplying unvalidated token IDs. Attackers can send integer-array prompts to the llama-server /completion endpoint with IDs outside 0 to n_vocab to read embedding memory out of bounds or crash the process. | Tue 27 Oct |
| CVE-2026-116816 | 8.5 high | 0.00% | no CPE yet | llama.cpp b5899 before b10261 contains a heap out-of-bounds write vulnerability in llm_tokenizer_plamo2::build that allows attackers to corrupt heap memory via crafted GGUF models. Attackers can supply BYTE-typed token text like <0x-1> whose negative parsed value writes token ids before the bytes_ buffer during model loading, causing crashes or potential code execution. | Tue 27 Oct |
| CVE-2026-43622 | 8.5 high | 0.00% | ggml, llama.cpp | llama.cpp builds b1886 through b7445 contain a double free vulnerability in the LLaMA-Android JNI wrapper where new_1batch() allocates memory using malloc() while free_1batch() deallocates it using the C++ delete operator, causing heap metadata corruption. Attackers can trigger this memory management mismatch to cause denial of service through process crashes or potentially achieve arbitrary code execution depending on allocator state. | Fri 23 Oct |
| CVE-2026-43627 | 8.5 high | 0.00% | ggml, llama.cpp | llama.cpp builds b1283 through b9058 contain an integer overflow vulnerability in the llama_batch_init() function where unchecked multiplications in malloc() calls can wrap past INT32_MAX when computing allocation sizes. Attackers can pass specially crafted parameters to trigger integer overflow, causing heap corruption and potentially achieving arbitrary code execution through subsequent batch operations that write past allocated buffer boundaries. | Fri 23 Oct |
| CVE-2026-43628 | 8.5 high | 0.00% | ggml, llama.cpp | llama.cpp builds b3978 through b9058 contain an integer underflow and out-of-bounds read vulnerability in the DRY sampler that allows unauthenticated attackers to trigger a heap buffer underflow by sending a crafted HTTP request with dry_allowed_length set to INT32_MIN to the /v1/completions or /v1/chat/completions endpoints. Attackers can exploit this vulnerability to crash the server with SIGSEGV causing denial of service for all connected users, or corrupt token sampling probabilities by reading garbage values from memory before the allocated buffer. | Fri 23 Oct |
| CVE-2026-70640 | 7.3 high | 0.00% | ggml, llama.cpp | llama.cpp builds b1886 through b7445 contain a race condition use-after-free vulnerability in the LLaMA-Android JNI wrapper where bench_1model() and free_1context() lack synchronization, allowing Thread A to operate on freed memory while Thread B concurrently frees the llama_context. Attackers can exploit this by performing heap spray with attacker-controlled data containing a fake vtable to hijack the vtable pointer at offset +0x30, causing llama_batch_allocr::clear() to dereference arbitrary memory and achieve remote code execution. | Sat 24 Oct |
| CVE-2026-116840 | 6.9 medium | 0.00% | no CPE yet | llama.cpp b5854 before b11275 contains an integer overflow vulnerability in gguf_init_from_reader that bypasses the CVE-2025-53630 size check when padding near-SIZE_MAX tensor sizes. Attackers can supply a crafted GGUF file whose tensor size pads to zero, causing an undersized data blob and out-of-bounds reads by consumers. | Tue 27 Oct |
| CVE-2026-116829 | 6.9 medium | 0.00% | no CPE yet | llama.cpp b10398 before b11301 contains an integer overflow vulnerability in gguf_init_from_reader that allows attackers to bypass tensor element-count validation by supplying shapes whose product wraps to zero. Attackers can craft GGUF files with enormous tensor extents that overflow ggml_nbytes() and strides, causing out-of-bounds memory access when buffers are sized from wrapped byte counts. | Tue 27 Oct |
| CVE-2024-58399 | 6.9 medium | 0.00% | no CPE yet | llama.cpp before b2731 contains an out-of-bounds read vulnerability in llama_model_loader that allows attackers to read memory outside the mapped model file via crafted GGUF tensor offsets. Attackers can supply a crafted or truncated GGUF model file with arbitrary 64-bit tensor offsets to crash the process or load adjacent process memory as model weights. | Tue 27 Oct |
| CVE-2025-71439 | 6.8 medium | 0.00% | no CPE yet | llama.cpp before b5649 contains a heap out-of-bounds read vulnerability in llama_vocab::impl::print_info() that allows attackers to crash model loading via oversized special token IDs. Attackers can supply a crafted GGUF model with a short tokenizer.ggml.tokens array, causing llama-cli or llama-server to read past id_to_token and crash. | 8 h ago |
| CVE-2026-70639 | 6.8 medium | 0.00% | ggml, llama.cpp | llama.cpp builds b1886 through b7445 contain a null pointer dereference vulnerability in the LLaMA-Android JNI wrapper where the bench_1model() function fails to validate the model context pointer before dereferencing it. Attackers can supply a malicious, corrupt, or truncated model file to trigger a null context condition, causing a SIGSEGV crash that terminates the Android application process and results in denial of service. | Sat 24 Oct |
| CVE-2026-43630 | 6.3 medium | 0.00% | ggml, llama.cpp | llama.cpp builds b4882 before b8487 contain an out-of-bounds read vulnerability in the recurrent memory state restore path that allows attackers with write access to the slot save directory to read memory past the end of the allocated cells array. Attackers can craft a malicious slot file with an oversized seq_id value to trigger an out-of-bounds read that leaks heap data including pointer values into server logs, defeating ASLR protections and facilitating further exploitation. | Thu 22 Oct |
| CVE-2026-116096 | 1.9 low | 0.00% | no CPE yet | A weakness has been identified in ggml-org llama.cpp. Affected by this issue is the function list_pockettts_state_slots of the file tools/mtmd/models/pockettts-gen.cpp of the component Pocket-TTS. This manipulation of the argument mmproj causes out-of-bounds write. The attack requires local access. The exploit has been made available to the public and could be used for attacks. Upgrading to version b11796 can resolve this issue. Patch name: 0a1ef309cdbb9f2024f58a38fb4c42352c5db569. The affected component should be upgraded. The reported heap out‑of‑bounds write (gen‑audio / pocket‑tts state_out path) is no longer a silent exploitable heap write on current master and the latest release b11871 - a malicious mmproj that previously triggered it now trips a release‑active GGML_ASSERT and aborts. But the submitter's recommended by‑construction fix (PR #29360 "prevent OOB accesses from malformed mmproj files") is still open. So the root cause is detected‑and‑aborted, not prevented‑by‑design. | Mon 26 Oct |
| CVE-2026-18581 | 1.9 low | 0.00% | no CPE yet | A vulnerability was determined in ggml-org llama.cpp e15efe0. Affected by this issue is some unknown functionality of the file common/jinja/parser.cpp of the component Jinja Minja Template Parser. Executing a manipulation with the input {{9|9|{ can lead to reachable assertion. The attack requires local access. The exploit has been publicly disclosed and may be utilized. The project was informed of the problem early through an issue report but has not responded yet. | Fri 23 Oct |
| CVE-2026-97620 | unscored | 0.00% | no CPE yet | In the Linux kernel, the following vulnerability has been resolved: drm/xe: Flush LSC untyped L1 dataport cache after rcs/ccs batches emit_render_cache_flush() sets PIPE_CONTROL0_HDC_PIPELINE_FLUSH to flush the L2/HDC data cache before fence signalling, but it never requests a flush of the LSC untyped L1 data cache via the 'Untyped Data-Port Cache Flush Enable' bit in PIPE_CONTROL DWord0[11]. Per the Bspec, in 3D pipeline mode HDC Pipeline Flush is documented to also flush/invalidate the untyped L1 cache, but only depending on how HDC_CHICKEN0[13:11] is programmed. Starting with MTL, this coupling between HDC Pipeline Flush and the untyped L1 cache flush no longer holds in practice, regardless of how HDC_CHICKEN0 is programmed, so relying on it is not safe on newer platforms such as BMG. Mesa's Vulkan driver (anv) has been assuming the kernel flushes both caches between submissions, and hit user-visible corruption in apps such as Llama.cpp because of this gap; it now works around it by flushing both caches again from userspace at the end of every command buffer. Correctness between submissions on the same queue is userspace's responsibility and belongs in Mesa, not the kernel. However, for security we must ensure stale data can't leak through the untyped L1 dataport cache once memory is reclaimed or evicted, which requires the KMD to flush it before releasing memory for reuse. Prior to MTL, HDC_CHICKEN0 could be programmed (as already done for DG2 via Wa_22010960976/Wa_14013347512) to reliably keep HDC Pipeline Flush coupled to the untyped L1 cache flush, so those platforms are unaffected. Mesa's own anv driver found that on MTL the HW disconnected the two independently of how HDC_CHICKEN0 is programmed, and could not bring the old behavior back even by writing the register by hand; see Mesa commit 7c2ff46a4fc3 ("anv: don't prevent L1 untyped cache flush in 3D mode"). The kernel can't reliably request the flush from the CS on MTL either, so restrict the new PIPE_CONTROL bit to GRAPHICS_VERx100 >= 2000 (Xe2 and later), where it can be relied on. Explicitly set PIPE_CONTROL0_UNTYPED_DATAPORT_CACHE_FLUSH together with PIPE_CONTROL0_HDC_PIPELINE_FLUSH in emit_render_cache_flush() on Xe2 and later, so the L1 data cache is known clean before memory is released for reuse, without depending on undocumented platform-specific HDC_CHICKEN0 behavior. Bspec: 56551 (cherry picked from commit 434514b6fe731e873808297c268fc52cdf4a1ce6) | 19 Oct |
26 of 26
Download these 26 as CSV · subscribe to this list.
EPSS scores: 2026-10-27. NVD cursor: 2026-10-28 21:20. The News page shows only notable CVEs (KEV, your stack, new criticals, EPSS over 50%, or active exploitation per SSVC); this table has all of them.