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.
OpenClaw v2026.2.24 adds multilingual stop keywords, native Android onboarding, ElevenLabs gateway config, and a multi-user security heuristic.
└──▷ GET THIS VERSION
$ git clone --branch v2026.2.24 https://github.com/openclaw/openclaw.git
# already have the repo? check out this version:$ git checkout v2026.2.24
›Adds multilingual auto-reply stop keywords (ES/FR/ZH/HI/AR/JP/DE/PT/RU) plus trailing-punctuation tolerance and exact do not do that as a stop trigger.
›Ships a native four-step Android onboarding flow and a five-tab shell (Connect, Chat, Voice, Screen, Settings) with a full Connect setup/manual mode screen.
›Adds provider-agnostic Talk configuration with legacy compatibility and exposes ElevenLabs gateway Talk config metadata for setup/status surfaces.
›Introduces security.trust_model.multi_user_heuristic to flag likely shared-user ingress, with hardening guidance for multi-user deployments (sandbox.mode="all", workspace-scoped FS, reduced tool surface).
└──▷ BREAKING ON UPGRADE
!Heartbeat delivery now blocks direct/DM targets (e.g. user:<id>, Telegram user chat IDs, WhatsApp direct numbers/JIDs); heartbeat runs still execute but direct-message delivery is skipped — only channel/group targets receive outbound heartbeat messages.
!Docker network: "container:<id>" namespace-join mode is blocked by default for sandbox and sandbox-browser containers; to re-enable it, set agents.defaults.sandbox.docker.dangerouslyAllowContainerNamespaceJoin: true.
Lightweight coding agent that runs in your terminal
Codex CLI v0.105.0 adds syntax-highlighted diffs, voice dictation, multi-agent fan-out from CSV, new TUI commands, and granular approval controls.
└──▷ GET THIS VERSION
$ git clone --branch rust-v0.105.0 https://github.com/openai/codex.git
# already have the repo? check out this version:$ git checkout rust-v0.105.0
└──▷ USE IT
Enable voice dictation so you can hold spacebar to record and transcribe prompts instead of typing.
toml
# In your Codex config file (e.g. ~/.codex/config.toml)
[features]
voice_transcription = true
›Adds syntax highlighting for fenced code blocks and diffs in the TUI, plus a /theme picker with live preview and theme-aware diff colors for light and dark terminals.
›Enables voice dictation in the TUI: hold spacebar to record and transcribe a prompt directly (enable with features.voice_transcription = true in config).
›New spawn_agents_on_csv capability fans out multi-agent work from a CSV file with built-in progress and ETA tracking.
›Sub-agents now support nicknames, a cleaner picker UI, and visible child-thread approval prompts for easier multi-agent tracking.
›New /copy TUI command copies the latest complete assistant reply to the clipboard.
+15 moreshow less
›New /clear command and Ctrl-L shortcut clear the screen without losing thread context; /clear can also start a fresh chat.
›Codex can now request extra sandbox permissions for a command mid-session, and specific approval prompt types can be auto-rejected without disabling approvals entirely.
›App-server thread/list now supports searching threads by title.
›Thread status is now exposed in thread/read and thread/list responses and in notifications.
›App-server thread/resume returns the latest turn inline, reducing data loss on reconnects.
›Adds configurable MCP OAuth callback URL for MCP server login.
›Adds allow_login_shell config option.
›Implements proxy-only egress for the Linux sandbox via a TCP-UDS-TCP bridge.
›Node module resolution paths can now be specified for js_repl.
›Adds configurable agent spawn depth.
›Adds configurable write_stdin timeout.
›Adds network proxy config wiring via permissions.network proxy config.
›Adds JSON tracing logs for the app-server.
›Adds ability to attach extra files to feedback submissions.
›Profile-level model_catalog_json override is now supported.
An open-source AI agent that brings the power of Gemini directly into your terminal.
gemini-cli v0.30.0 adds a --policy flag, Ctrl-Z suspension, tool output masking, multi-line ask-user input, strict seatbelt profiles, and an initial SDK package.
└──▷ GET THIS VERSION
$ git clone --branch v0.30.0 https://github.com/google-gemini/gemini-cli.git
# already have the repo? check out this version:$ git checkout v0.30.0
└──▷ TRY IT
Enforce a custom tool policy file on every invocation instead of using the deprecated --allowed-tools flag.
$ gemini --policy ./my-policy.json
Suspend a long-running gemini-cli session to the background to run another shell command, then resume.
$ # Press Ctrl-Z inside an active gemini-cli session to suspend it
# Then resume with:
fg
›Adds --policy flag for user-defined tool policies, replacing --allowed-tools and excludeTools.
›Enables tool output masking by default to reduce sensitive data exposure in tool responses.
›Supports multi-line text answers in the ask-user tool for richer interactive prompts.
›Adds Ctrl-Z process suspension support for pausing and backgrounding the CLI.
›Introduces strict seatbelt profiles for tighter sandbox enforcement.
+15 moreshow less
›Adds autoconfigure memory usage setting directly in the settings dialog.
›Adds a setting to hide the shortcuts hint UI.
›Introduces limits on search results to control context size.
›Formalizes a 5-phase sequential planning workflow in Plan Mode.
›Allows skills to be enabled in Plan Mode.
›Updates internal utility models to Gemini 3.
›Adds A2A value-resolver for auth credential resolution.
›Enables 30-day session retention by default.
›Adds text wrapping support for markdown tables in the terminal UI.
›Adds a searchable list UI backing settings and extensions panels.
›Shows a notification when an extension command conflicts with a built-in command.
›Enables in-CLI extension management commands.
›Bootstraps an initial SDK package with SessionContext support, dynamic system instructions, and custom skills.
›Fallback to chat-base model automatically when an unrecognized model is specified for chat.
›Adds admin controls documentation and policy overrides for Plan Mode.
└──▷ BREAKING ON UPGRADE
!--allowed-tools and excludeTools are deprecated in favor of the policy engine; existing configurations relying on these flags will need to migrate to the --policy flag.
llama.cpp server gains multi-modal context checkpoint support for hybrid and recurrent memory architectures.
└──▷ GET THIS VERSION
$ git clone --branch b8152 https://github.com/ggml-org/llama.cpp.git
# already have the repo? check out this version:$ git checkout b8152
›Adds multi-modal context checkpoint support to the server, enabling checkpoint/restore of KV cache state across hybrid-ISWA and recurrent memory backends.
›Adds --disable-access-log-for-endpoints CLI flag to suppress access logging for specific endpoints.
›Adds GGUF repo_id:quant_type syntax for selecting quantization variants when loading GGUF models.
›Adds nested config support in YAML files for vLLM server configuration.
›Adds FlashInfer autotune control flag (#34006) to manage FlashInfer autotuning behavior at runtime.
›Adds native NCCL-based weight syncing API for RLHF workflows, enabling direct weight updates between training and inference engines without third-party communication layers.
+31 moreshow less
›Adds engine pause/resume with request preservation for RLHF workflows, allowing training loops to halt inference without dropping in-flight requests.
›Adds layerwise weight reloading for QeRL in RLHF workflows.
›Adds WebSocket-based Realtime API for streaming audio interactions, including Voxtral realtime support.
›Adds sampling parameters and token/prompt token ID return fields to the Responses API.
›Adds /server_info collect_env endpoint for retrieving environment diagnostics from a running server.
›Adds OTEL tracing during model loading for observability of startup phases.
›Adds run_batch transcription and translation support.
›Adds HF Hub LoRA resolver (#20320) for loading LoRA adapters directly from Hugging Face Hub.
›Adds KV offloading connector metrics and labeled prompt token metrics for P/D disaggregation.
›Adds Unified Parallel Drafting for speculative decoding, delivering throughput improvements alongside structured output support and penalty application in Model Runner V2.
›Adds async scheduling with Pipeline Parallelism, delivering 30.8% end-to-end throughput improvement and 31.8% TPOT improvement.
›Adds FP8 block quant for CompressedTensorsW8A16Fp8, ModelOpt MXFP8 for dense models, and NVFP4/FP8 support on Turing GPUs.
›Adds TP > 4 support for FP4 Gemm quantization.
›Adds Mooncake disaggregated serving connector rework with bootstrap server and cross-layer KV cache layout at NIXL Connector V2.
›Adds EPLB logical expert capture with router replay for expert-parallel load balancing.
›Adds MoE cold start optimization option for torch.compile to reduce compilation overhead.
›Adds Intel XPU support for unquantized MoE, MXFP4 MoE, WNA16, scaled_mm, and FP8 MoE kernels via vllm-xpu-kernels (replacing IPEX).
›Adds ARM CPU KleidiAI INT4 dynamic quantization with BF16 activations and NEON BFMMLA BF16 paged attention.
›Adds IBM Z BF16 kernel type for s390x architecture.
›Adds SM100 INT4 W4A16 kernel and SM121 (DGX Spark) CUTLASS support for NVIDIA GPUs.
›Adds MNNVL protocol support for GB series NVIDIA hardware.
›Adds ORJSONResponse for faster API response serialization.
›Adds DeepSeek ReasoningParser with thinking enabled by default.
›Adds early tokenization validation to surface malformed inputs before scheduling.
›Adds LoRA expansion for Gemma3 vision components, Nemotron-H MTP models, and Qwen3 output embedding.
›Adds new model architecture support for GLM-OCR with MTP, Qwen3-ASR, DeepSeek-OCR-2, Intern-S1-Pro, MiniCPM-o 4.5, openPangu7B-VL, NemotronHPuzzle heterogeneous, MusicFlamingo, FunAudioChat, ColBERT late interaction, voyage-4-nano, and GLM-5.
›Adds EAGLE3 speculative decoding support for Hunyuan/HunyuanVL, AFMoE, and Mistral3.
›Adds Pooling API request schema consensus for ScoreRequest and final standardization.
›Adds PluggableLayer support applied to linear layers and Mamba layers.
›Only includes the Authorization header in outbound requests when OPENAI_API_KEY is set, reducing credential leakage risk.
└──▷ BREAKING ON UPGRADE
!Removed BitBlas quantization backend — workloads using BitBlas will fail after upgrade.
!Removed Marlin 24 quantization backend — workloads using Marlin 24 will fail after upgrade.
!Removed deprecated reasoning_content message field — code referencing this field will break.
!Removed deprecated pooling items — any configuration or API calls relying on the deprecated pooling endpoints/fields will break.
!Removed deprecated VLLM_ALL2ALL_BACKEND environment variable — scripts setting this variable will have no effect (and previously working behavior dependent on it is gone).
!Deprecated IPEX for Intel XPU; switched to vllm-xpu-kernels — existing XPU deployments using IPEX will need to migrate to vllm-xpu-kernels.
LanceDB python-v0.30.0-beta.2 adds parallel local inserts and a type-safe Rust expression builder API for query filters.
└──▷ GET THIS VERSION
$ git clone --branch python-v0.30.0-beta.2 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout python-v0.30.0-beta.2
›Adds an expression builder API in the Rust client for constructing type-safe query filters programmatically.
›Enables parallel inserts for local tables, improving write throughput for local LanceDB deployments.
›Upgrades the Node.js bindings from napi-rs v2 to v3.
›Hooks up a new writer backend for insert operations.
LanceDB v0.27.0-beta.2 adds a Rust expression builder API for type-safe query filters and parallel inserts for local tables.
└──▷ GET THIS VERSION
$ git clone --branch v0.27.0-beta.2 https://github.com/lancedb/lancedb.git
# already have the repo? check out this version:$ git checkout v0.27.0-beta.2
›Adds expression builder API in the Rust SDK for constructing type-safe query filters programmatically.
›Enables parallel inserts for local tables, improving write throughput.
›Upgrades napi-rs from v2 to v3 in the Node.js bindings, bringing the latest NAPI runtime support.