The daily firehose — everything the toolchain shipped today, already filtered.
// HOW THIS ISSUE IS MADE
We read every release from the 174 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.
Agno v2.3.5 adds OpenTelemetry-based native tracing and non-blocking background task hooks for agents and teams.
└──▷ GET THIS VERSION
$ git clone --branch v2.3.5 https://github.com/agno-agi/agno.git
# already have the repo? check out this version:$ git checkout v2.3.5
›Introduces OpenTelemetry-based native tracing that automatically captures and stores agent runs, model calls, tool executions, and team operations in your Agno database.
›Agent and Team pre- and post-hooks can now run as background tasks on AgentOS for fully non-blocking, concurrent execution — useful for notifications, logging, or evaluations not on the critical path.
›Adds a debug-level environment variable for controlling Agno debug output.
›Extends debug-level support to workflows.
›Unifies model authentication errors across providers into a consistent error surface.
AutoGPT Platform adds Human-in-the-Loop blocks, Safe Mode toggle, external API routes, and a Discord thread block.
└──▷ GET THIS VERSION
$ git clone --branch autogpt-platform-beta-v0.6.39 https://github.com/Significant-Gravitas/AutoGPT.git
# already have the repo? check out this version:$ git checkout autogpt-platform-beta-v0.6.39
›Adds external API routes for store search and tool usage, plus integration management via external API.
›Adds a Human In The Loop block with a review workflow, letting agents pause for human approval mid-execution.
›Implements a graph-level Safe Mode toggle to control which HITL blocks require human review.
›Adds a Google Drive Picker field type for enhanced file selection in graph inputs.
›Adds a Discord 'create thread' block for automating Discord thread creation from flows.
+2 moreshow less
›Adds a trigger agent banner in the UI for webhook-based flows.
›Enhances BlockMenuSearch to support adding agents directly from the block menu.
$ git clone --branch dotnet-1.68.0 https://github.com/microsoft/semantic-kernel.git
# already have the repo? check out this version:$ git checkout dotnet-1.68.0
└──▷ USE IT
Access Gemini's chain-of-thought reasoning content alongside the final response.
csharp
var executionSettings = new GeminiPromptExecutionSettings
{
IncludeThoughts = true
};
var result = await chatService.GetChatMessageContentsAsync(history, executionSettings, kernel);
Tune reasoning depth for Gemini 3.0+ models by setting ThinkingLevel in GeminiThinkingConfig.
csharp
var executionSettings = new GeminiPromptExecutionSettings
{
ThinkingConfig = new GeminiThinkingConfig
{
ThinkingLevel = ThinkingLevel.High
}
};
var result = await chatService.GetChatMessageContentsAsync(history, executionSettings, kernel);
›Adds IncludeThoughts parameter to the Google Connector to expose Gemini reasoning/chain-of-thought content in responses.
›Adds ThinkingLevel parameter to GeminiThinkingConfig for controlling reasoning depth on Gemini 3.0+ models.
›Adds support for custom ToolChoice and ParallelToolCallsEnabled options in OpenAIResponseAgent.
›Enables argument type retention in handoff orchestration.
›Upgrades target framework to .NET 10.
+1 moreshow less
›Adds an OpenAPI tool call migration sample for AzureOpenAI.
›Supports forking/rebranding the CLI by setting piConfig.name and piConfig.configDir in package.json, updating the banner, help text, config paths, and error messages without code changes.
llama.cpp b7256 extends CUDA Flash Attention to Volta GPUs via generalized MMA kernel support.
└──▷ GET THIS VERSION
$ git clone --branch b7256 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b7256
›Adds Volta GPU architecture support to the CUDA Flash Attention (FA) implementation via a generalized matrix multiply-accumulate (MMA) kernel, expanding hardware compatibility for CUDA-accelerated inference.
└──▷ BREAKING ON UPGRADE
!Linux releases will soon switch from .zip archives to .tar.gz archives — deployment scripts that fetch or unpack Linux release artifacts will need to be updated.
llama.cpp b7247 adds WebGPU backend support for Emscripten/Wasm builds with fast matrix multiply and Q4_0 quantization.
└──▷ GET THIS VERSION
$ git clone --branch b7247 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b7247
›Adds WebGPU backend support for Emscripten (Wasm) builds, enabling browser-side inference with fast f16/f32 matrix and matrix-vector multiplication via subgroup matrix shaders.
›Supports Q4_0 quantized matrix multiply in the WebGPU backend, bringing quantized inference to Wasm targets.
›Adds memory64 support via Emscripten's memory64 option to enable get_memory in Wasm builds.
›Enables subgroup matrix (cooperative matrix) acceleration in the WebGPU backend, with automatic detection of available subgroup matrix configurations at runtime.
›Adds toggles for subgroup matrix and f16 support specifically on Nvidia+Vulkan targets.
└──▷ BREAKING ON UPGRADE
!Linux releases will switch from .zip archives to .tar.gz archives — deployment scripts that unzip Linux release artifacts will need to be updated.
vLLM v0.12.0 adds GPU Model Runner V2, EAGLE multimodal speculative decoding, RLHF pause/resume, new optimization levels, and major quantization/API expansions.
└──▷ GET THIS VERSION
$ git clone --branch v0.12.0 https://github.com/vllm-project/vllm.git
# already have the repo? check out this version:$ git checkout v0.12.0
└──▷ TRY IT
Start vLLM with aggressive compilation to maximize throughput at the cost of longer startup, useful for long-running production deployments.
$ vllm serve meta-llama/Llama-3-8B-Instruct -O3
Serve a quantized GGUF model directly from HuggingFace using the new repo_id:quant_type syntax without manually downloading files.
›Adds -O0, -O1, -O2, -O3 optimization-level flags to trade startup time for runtime performance, with more compilation flags planned.
›Adds --fully-sharded-loras flag enabling FusedMoE LoRA support.
›Moves flat logprob control from an environment variable into SamplingParams, making it a first-class sampling parameter.
›Adds verbose_json and timestamp features to the Whisper transcription/translation API.
›Adds parallel_tool_calls param compliance and parsed tool arguments support to the tool-calling API.
+27 moreshow less
›Adds inputs_embeds_size config key as a separate dimension from hidden_size for flexible embedding configuration.
›Supports GGUF model loading from HuggingFace using repo_id:quant_type syntax for improved UX.
›Adds parallel_tool_calls param and tool filtering support in ToolServer.
›Adds Responses API multi-turn support for non-harmony requests, plus reasoning item input parsing.
›Adds GPU Model Runner V2 (experimental): GPU-persistent block tables, Triton-native sampler with per-request seeds, memory-efficient prompt logprobs, and simplified DP/CUDA graph implementations.
›Adds Prefill Context Parallel (PCP) to partition the sequence dimension during prefill, complementing existing Decode Context Parallel (DCP) for improved long-sequence inference.
›Adds RLHF pause and resume generation support for asynchronous RL training workflows.
›Adds cross-layer KV blocks support and KV cache residency metrics.
›Adds audio embeddings support in chat completions.
›Adds EAGLE speculative decoding with multi-step CUDA graph support, DP>1, and multimodal support including Qwen3VL.
›Adds logprobs support with speculative decoding and async scheduling.
›Adds iteration-level profiling for Torch and CUDA profiler.
›Adds colorized log output.
›Adds new model families: PLaMo-3, OpenCUA-7B, HunyuanOCR, Mistral Large 3, and Ministral 3.
›Adds Gemma3 GGUF multimodal format support.
›Adds Qwen3 Omni audio-in-video multimodal support.
›Delivers 18.1% throughput improvement and 10.7% TTFT improvement on DeepSeek-V3.1 via batch invariant BMM optimization.
›Delivers 2.2% throughput improvement and 3.6% TTFT improvement at batch size 32 via shared experts overlap with FlashInfer DeepGEMM.
›Adds W4A8 Marlin kernel quantization support.
›Adds NVFP4 MoE CUTLASS support for SM120 and TRTLLM MoE NVFP4 kernel.
›Adds AWQ compressed-tensors support for Turing GPUs.
›Adds FusedMoE LoRA Triton kernel for MXFP4 quantization.
›Adds AMD ROCm support for DeepSeek v3.2 with SparseMLA, FP8 MLA decode, AITER attention backend, bitsandbytes quantization on warp-size-32 GPUs, and sliding window for AiterFlashAttentionBackend.
›Adds paged attention GEMM acceleration on ARM CPUs with NEON.
›Adds FlashAttention ViT support, now the default attention backend.
›Adds enhanced NUMA functionality for systems with multiple NUMA nodes per socket.
›Reduces Docker image size by approximately 200MB.
└──▷ BREAKING ON UPGRADE
!PyTorch upgraded to 2.9.0 with CUDA 12.9 — environments must be updated to CUDA 12.9.
!The num_lookahead_slots parameter has been removed.
!The best_of parameter has been removed.
!LoRA extra vocab has been removed.
!The xformers backend is deprecated.
!Mistral format is now auto-detected during model loading, which may change behavior for existing Mistral model configs.
$ git clone --branch v3.136.0 https://github.com/langfuse/langfuse.git
# already have the repo? check out this version:$ git checkout v3.136.0
›Adds environment variable LANGFUSE_TRACE_DELETE_SKIP_PROJECT_IDS to drain specific deletions for a specified project, giving operators fine-grained control over trace deletion queues.
›Extends the LLM Connections API to allow setting provider-specific config for Bedrock and VertexAI integrations.
›Supports setting a custom Keycloak provider name via the auth configuration.
›Adds CANCELLED status to BatchExportStatus and handles cancellation during job processing.
›Maps llm.input_messages and llm.output_messages OpenTelemetry semantic conventions for OTEL-based ingestion.
+2 moreshow less
›Adds a filter sidebar to the LLM-as-a-judge eval table for narrowing evaluation results.
›Supports setting a custom Google AI base URL in LLM connection configuration.
THE AI TOOLCHAIN