> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fortytwo.network/llms.txt
> Use this file to discover all available pages before exploring further.

# Changelog

<Update label="[|] Fortytwo P2P Relay 0.6.0" description="2026-04-07" tags={["[|] Fortytwo P2P Relay"]}>
  ### Fixed

  * Autonat v2 never determining NAT status: calling `add_external_address()` in the identify handler marked the observed address as confirmed before identify could emit `NewExternalAddrCandidate`, causing the swarm to suppress the candidate event. Autonat v2 client never received candidates to test, leaving NAT status unknown and triggering infinite node restarts.
</Update>

<Update label="[|] Fortytwo P2P Relay 0.5.0" description="2026-04-02" tags={["[|] Fortytwo P2P Relay"]}>
  ### Fixed

  * Gossipsub mesh shrinkage has been fixed to adapt to traffic rates. Positive per-topic weights are now applied.
  * Non-globally routable addresses are now filtered before emitting `NewExternalAddrOfPeer` events. This prevents loopback and private addresses from entering the peerstore and being dialed by Gossipsub and other protocols.
</Update>

<Update label="⏃ Protocol 0.25.1" description="2026-04-01" tags={["⏃ Protocol"]}>
  ### Changed

  * Increased the limit for pending incoming connections from 64 to 256.
</Update>

<Update label="⏃ Protocol 0.25.0" description="2026-03-26" tags={["⏃ Protocol"]}>
  ### Added

  * `ConnectionLimits` behaviour to bound per-peer (default 2) and pending incoming (default 64) connections, preventing unbounded connection accumulation from misbehaving peers.
  * New `NetworkLimitsConfig` with env-configurable fields: `FT_YAMUX_MAX_STREAMS`, `FT_MAX_ESTABLISHED_PER_PEER`, `FT_MAX_PENDING_INCOMING`.

  ### Changed

  * Upgraded AutoNAT from v1 to v2 for wire compatibility with bootstrap nodes (v2 server).

  ### Fixed

  * DCUtR failure logs now include attempt count (e.g. `attempt 1/3`) so operators can distinguish transient first-attempt failures from actual problems at max retries.
  * Fixed yamux substream asymmetry: relay client transport now uses the same yamux config as direct transport instead of the 256-stream default.
  * Lowered default `yamux_max_streams` from 1024 to 256 (sufficient for `mesh_n_high=2`).
</Update>

<Update label="[|] Fortytwo P2P Relay 0.4.0" description="2026-03-26" tags={["[|] Fortytwo P2P Relay"]}>
  *General stability improvements.*
</Update>

<Update label="⏃ Protocol 0.24.0" description="2026-03-02" tags={["⏃ Protocol"]}>
  ### Fixed

  * Only confirmed external addresses are now advertised to peers.
  * Preventing non-routable addresses from entering the DHT at swarm construction time.
</Update>

<Update label="።‖ Fortytwo P2P App 0.3.4" description="2026-02-25" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Status bar showing current system load,
  * Status bar also shows an overlay for active downloads.
  * Per-model performance stats: Win Rate, Inference Participation Rate, and Tokens per Hour with detailed on-hover tooltips.
  * Reward Summary now shows rewards confirmed on the blockchain, as well as pending and lost rounds which are tracked locally by the App.

  ### Changed

  * Improved model storage and cleanup to better protect manually selected models.
  * Smarter auto-model cleanup to reduce disk usage without breaking active setups.
  * Rewards and rounds are now updated reliably, even when the node is paused or stopped.
  * Improved "Start after download" and "Enable when ready" behaviors (the latter is now enabled by default).
  * Updated colors and text styles for both dark and light themes.

  ### Fixed

  * Fixed an issue where a model briefly appeared as "failed" after canceling a download.
  * Improved handling of a stuck capsule — it is now terminated on node startup.
</Update>

<Update label="⏃ Protocol 0.23.0" description="2025-02-23" tags={["⏃ Protocol"]}>
  ### Added

  * Relay circuit addresses are now propagated to connected peers via the Identify protocol, improving address discovery for nodes behind NATs.
  * Experimental infrastructure for direct peer-to-peer connectivity (DCUtR). Protocol nodes now listen on relay circuit addresses, laying the groundwork for NAT traversal via relay-assisted hole-punching. This feature is enabled by default but should be considered experimental — we will share production results in a future release.

  ### Improved

  * Relay connectivity is now always-on rather than requiring explicit activation.
  * Streamlined internal event handling for relay connections.
  * Tuned consensus message parameters to improve network stability.

  ### Fixed

  * Resolved an issue where Docker bridge IPs and other private network addresses could propagate to the DHT, causing `WrongPeerId` errors when peers attempted to dial.
  * Private and non-routable addresses are now filtered at multiple stages: peer store, Kademlia, and address advertisement.
  * Updated core networking dependencies for improved protocol compatibility.
</Update>

<Update label="[|] Fortytwo P2P Relay 0.3.0" description="2026-02-23" tags={["[|] Fortytwo P2P Relay"]}>
  ### Added

  * Rate limiting for relay reservations and circuits. Relays now enforce per-peer and per-IP token-bucket rate limits to prevent resource exhaustion from misbehaving or compromised peers.
  * Denial metrics. Relay operators can now see accepted/denied reservation and circuit counts (with denial reasons) in Grafana dashboards.

  ### Changed

  * Lowered relay resource caps to safer defaults: 128 max reservations (was 1024), 16 max circuits (was 32), with tighter per-peer limits.
</Update>

<Update label="⏃ Protocol 0.22.0" description="2026-02-13" tags={["⏃ Protocol"]}>
  ### Added

  * Wired up dial backoff to record connection failures and prevent rapid reconnection attempts.

  ### Security

  * Capped Brotli decompressed output to 4 MiB to prevent decompression bomb OOM attacks.
</Update>

<Update label="[|] Fortytwo P2P Relay 0.2.0" description="2026-02-13" tags={["[|] Fortytwo P2P Relay"]}>
  ### Fixed

  * Fixed autonat healthcheck.
</Update>

<Update label="[|] Fortytwo P2P Relay 0.1.0" description="2026-02-12" tags={["[|] Fortytwo P2P Relay"]}>
  *The initial release of Fortytwo P2P Relay.*

  Refer to the '[Fortytwo P2P Relay Quick Start guide](/docs/relay-quick-start)' to learn how to run the Relay Node via the Docker setup.
</Update>

<Update label=">_ Fortytwo P2P CLI 2026.01.23.0" description="2026-01-23" tags={[">_ Fortytwo P2P CLI"]}>
  ### Added

  Added new featured model:

  * **GLM-4.7-Flash Q4**: Long-context reasoning at high efficiency, with steady logic, math, and coding across large inputs.<br />
    *REPO = unsloth/GLM-4.7-Flash-GGUF*<br />
    *MODEL NAME = GLM-4.7-Flash-Q4\_K\_M.gguf*

  ### Changed

  * New model takes place of an older one on position #8: **Qwen3-30B-A3B-Thinking-2507 Q4** -> **GLM-4.7-Flash Q4**

  ### Removed

  Removed an older featured model:

  * **Qwen3-30B-A3B-Thinking-2507 Q4**<br />
    *REPO = unsloth/Qwen3-30B-A3B-Thinking-2507-GGUF*<br />
    *MODEL NAME = Qwen3-30B-A3B-Thinking-2507-Q4\_K\_M.gguf*
</Update>

<Update label="⬢ Featured Models 3.0.0" description="2026-01-23" tags={["⬢ Featured Models"]}>
  ### Added

  Added new featured model:

  * **GLM-4.7-Flash Q4**: Long-context reasoning at high efficiency, with steady logic, math, and coding across large inputs.<br />
    *REPO = unsloth/GLM-4.7-Flash-GGUF*<br />
    *MODEL NAME = GLM-4.7-Flash-Q4\_K\_M.gguf*

  ### Changed

  * For CLI, new model takes place of an older one on position #8: **Qwen3-30B-A3B-Thinking-2507 Q4** -> **GLM-4.7-Flash Q4**

  ### Removed

  Removed an older featured model:

  * **Qwen3-30B-A3B-Thinking-2507 Q4**<br />
    *REPO = unsloth/Qwen3-30B-A3B-Thinking-2507-GGUF*<br />
    *MODEL NAME = Qwen3-30B-A3B-Thinking-2507-Q4\_K\_M.gguf*
</Update>

<Update label="⎔ Capsule 0.2.9" description="2026-01-23" tags={["⎔ Capsule"]}>
  ### Added

  * Added support for **GLM-4.7-Flash**.
</Update>

<Update label=">_ Fortytwo P2P CLI 2026.01.13.1" description="2026-01-13" tags={[">_ Fortytwo P2P CLI"]}>
  ### Fixed

  * Changed the model file name path: GLM-4.7-Q4\_K\_M-00001-of-00005.gguf -> Q4\_K\_M/GLM-4.7-Q4\_K\_M-00001-of-00005.gguf
</Update>

<Update label=">_ Fortytwo P2P CLI 2026.01.13.0" description="2026-01-13" tags={[">_ Fortytwo P2P CLI"]}>
  ### Changed

  * Changed the outdated model file name: LGAI-EXAONE\_EXAONE-4.0-32B-Q4\_K\_M.gguf -> EXAONE-4.0-32B-Q4\_K\_M.gguf
</Update>

<Update label="⬢ Featured Models 2.2.0" description="2026-01-12" tags={["⬢ Featured Models"]}>
  ### Fixed

  * Changed the model file name path: GLM-4.7-Q4\_K\_M-00001-of-00005.gguf -> Q4\_K\_M/GLM-4.7-Q4\_K\_M-00001-of-00005.gguf
</Update>

<Update label="⬢ Featured Models 2.1.0" description="2026-01-12" tags={["⬢ Featured Models"]}>
  ### Changed

  * Changed the outdated model file name: LGAI-EXAONE\_EXAONE-4.0-32B-Q4\_K\_M.gguf -> EXAONE-4.0-32B-Q4\_K\_M.gguf
</Update>

<Update label="።‖ Fortytwo P2P App 0.3.3" description="2025-12-31" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * New Wallet page with a clear balance summary.
  * Reward estimation is now available: you can preview your upcoming rewards using this feature.
  * Blockchain events are now tracked more reliably, ensuring rewards are reflected correctly, even if they arrive later

  ### Changed

  * Redesigned top toolbar with quick Wallet access.
  * Updated the "Slow" tag logic: it is now calculated based on total system memory rather than currently available memory.

  ### Fixed

  * Unused models are now handled more safely, reducing the risk of removing anything you might still need.
  * Opening notifications no longer resets your timeline — your full history remains visible and uninterrupted.
  * Fixed an issue where the Settings page didn’t show a notification when switching between Auto and Manual modes.
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.12.30.0" description="2025-12-30" tags={[">_ Fortytwo P2P CLI"]}>
  ### Added

  Added new featured models:

  * **Kimi-K2-Thinking**: Efficient Mixture of Experts reasoning; balanced accuracy, coherence, and efficiency across general and technical tasks.<br />
    *REPO = unsloth/Kimi-K2-Thinking-GGUF*<br />
    *MODEL NAME = Q4\_K\_M/Kimi-K2-Thinking-Q4\_K\_M-000XX-of-00013.gguf*
  * **MiniMax-M2.1**: Versatile generalist with standout coding, creative and complex reasoning skills.<br />
    *REPO = unsloth/GLM-4.7-GGUF*<br />
    *MODEL NAME = GLM-4.7-Q4\_K\_M-0000X-of-00005.gguf*
  * **Rnj-1 8B**: Long-context reasoning abilities for science and technical questions; strong capabilities in code and maths.<br />
    *REPO = bartowski/EssentialAI\_rnj-1-instruct-GGUF*<br />
    *MODEL NAME = EssentialAI\_rnj-1-instruct-Q4\_K\_M.gguf*
  * **Nanbeige4-3B-Thinking-2511**: Thoughtful step-by-step solver; excels at competition math, graduate science, and function calling.<br />
    *REPO = bartowski/Nanbeige\_Nanbeige4-3B-Thinking-2511-GGUF*<br />
    *MODEL NAME = Nanbeige\_Nanbeige4-3B-Thinking-2511-Q4\_K\_M.gguf*

  ### Changed

  * **Kimi-K2-Thinking** added to #3 position.
  * **MiniMax-M2.1** added to #4 position.
  * Models in EXTREME TIER, HEAVY TIER, and LIGHT TIER get +2 in numeration (e.g. 3 -> 5) up to #20 **Rnj-1-8B**.
  * **Rnj-1-8B** added to position #20 and all models after it have shifted by +1 in numeration, becoming +3 in total relatively to the past version order (e.g. 18 -> 21, 19 -> 22) up to #25 **Nanbeige4-3B-Thinking-2511**.
  * **Nanbeige4-3B-Thinking-2511** added to position #25 and all models after it have shifted by +1 in numeration, becoming +4 in total relatively to the past version order (e.g. 22 -> 26, 23 -> 27).

  ### Fixed

  * **Gemma-3 4B Q4** -> **Gemma-3 12B Q4**. The old name was incorrect and pointed to the 12B repository.
</Update>

<Update label="⬢ Featured Models 2.0.0" description="2025-12-30" tags={["⬢ Featured Models"]}>
  ### Added

  * **Kimi-K2-Thinking**: Efficient Mixture of Experts reasoning; balanced accuracy, coherence, and efficiency across general and technical tasks.<br />
    *REPO = unsloth/Kimi-K2-Thinking-GGUF*<br />
    *MODEL NAME = Q4\_K\_M/Kimi-K2-Thinking-Q4\_K\_M-000XX-of-00013.gguf*
  * **MiniMax-M2.1**: Versatile generalist with standout coding, creative and complex reasoning skills.<br />
    *REPO = unsloth/GLM-4.7-GGUF*<br />
    *MODEL NAME = GLM-4.7-Q4\_K\_M-0000X-of-00005.gguf*
  * **Rnj-1 8B**: Long-context reasoning abilities for science and technical questions; strong capabilities in code and maths.<br />
    *REPO = bartowski/EssentialAI\_rnj-1-instruct-GGUF*<br />
    *MODEL NAME = EssentialAI\_rnj-1-instruct-Q4\_K\_M.gguf*
  * **Nanbeige4-3B-Thinking-2511**: Thoughtful step-by-step solver; excels at competition math, graduate science, and function calling.<br />
    *REPO = bartowski/Nanbeige\_Nanbeige4-3B-Thinking-2511-GGUF*<br />
    *MODEL NAME = Nanbeige\_Nanbeige4-3B-Thinking-2511-Q4\_K\_M.gguf*

  ### Changed

  Model numbers for Fortytwo P2P CLI were updated:

  * **Kimi-K2-Thinking** added to #3 position.
  * **MiniMax-M2.1** added to #4 position.
  * Models in EXTREME TIER, HEAVY TIER, and LIGHT TIER get +2 in numeration (e.g. 3 -> 5) till #20 **Rnj-1-8B**.
  * **Rnj-1 8B** added to position #20 and all models after it have shifted by +1 in numeration, becoming +3 in total relatively to the past version order (e.g. 18 -> 21, 19 -> 22) till #25 **Nanbeige4-3B-Thinking-2511**.
  * **Nanbeige4-3B-Thinking-2511** added to position #25 and all models after it have shifted by +1 in numeration, becoming +4 in total relatively to the past version order (e.g. 22 -> 26, 23 -> 27).
</Update>

<Update label="⎔ Capsule 0.2.8" description="2025-12-29" tags={["⎔ Capsule"]}>
  ### Added

  * Added support for **Rnj-1 8B**, **GLM-4.7**, **Kimi-K2-Thinking**.
</Update>

<Update label="።‖ Fortytwo P2P App 0.3.2" description="2025-12-24" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Cache for reward and wallet values so the latest available data is shown on startup and after disconnects.

  ### Changed

  * **Improved model tags**: recommended models in manual mode are now calculated based on available system resources rather than total resources.
  * **Optimized RMB menu on system tray icon**:
    * Removed Color Scheme selection. It can now be defined in Settings from within the app.
    * Replaced Parameters with Settings, which opens the app directly to the Settings menu.

  ### Fixed

  * Visualization mode issues that occurred after disconnecting.
</Update>

<Update label="⏃ Protocol 0.21.0" description="2025-12-22" tags={["⏃ Protocol"]}>
  ### Fixed

  * Fixed round's leaderboard table.
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.12.19.0" description="2025-12-19" tags={[">_ Fortytwo P2P CLI"]}>
  ### Changed

  * (Windows) Resolved script execution warnings.
</Update>

<Update label="⊡ Fortytwo P2P Container 0.1.0" description="2025-12-18" tags={["⊡ Fortytwo P2P Container"]}>
  *The initial release of Fortytwo P2P Container.*

  Refer to the '[Fortytwo P2P Container Quick Start guide](/docs/quick-start)' to learn how to run the Inference Node via the Docker setup.
</Update>

<Update label="።‖ Fortytwo P2P App 0.3.1" description="2025-12-17" tags={["።‖ Fortytwo P2P App"]}>
  ### Changed

  * The contract address used to fetch the FOR token balance.

  ### Fixed

  * Resolved an issue preventing the p2p-known-peers file from being saved on certain systems.
</Update>

<Update label="⏃ Protocol 0.20.0" description="2025-12-17" tags={["⏃ Protocol"]}>
  ### Fixed

  * Bug fixes.
</Update>

<Update label="⏃ Protocol 0.19.0" description="2025-12-17" tags={["⏃ Protocol"]}>
  ### Added

  * Additional technical logs.

  ### Changed

  * Updated smart contract addresses.
  * Optimized p2p network.

  ### Fixed

  * Fixed bandwidth metrics.
  * Improved network load balancing.
</Update>

<Update label="።‖ Fortytwo P2P App 0.3.0" description="2025-12-15" tags={["።‖ Fortytwo P2P App"]}>
  *Introducing a major update with node operation visualization. We’ve added a timeline that shows the node’s recent round performance history and placed an emphasis on visualizing the round phases right above the timeline. Now, when your node participates in a round, you can see what it does: when it enters the round, generates a response, or evaluates responses from other participants. This is the same information you get from the Fortytwo P2P CLI logs, but presented in a clean UI with animations*

  ### Added

  * **Timeline view:** Node performance timeline visualization.
  * **Round status bar:** Animation of the current round status above the timeline.
  * **KV Cache “Apply” button:** After changing cache settings, click the Apply button—no need to restart the node manually.
  * **Transparent download status on startup:** The startup screen now shows download progress when an update is required.

  ### Changed

  * Improved network disconnection detection for more reliable and accurate handling.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.11" description="2025-12-11" tags={["።‖ Fortytwo P2P App"]}>
  ### Changed

  * A new Hugging Face API handle has been adopted for searching in the model manager.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.10" description="2025-12-09" tags={["።‖ Fortytwo P2P App"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.9" description="2025-12-08" tags={["።‖ Fortytwo P2P App"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.8" description="2025-12-05" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Settings page containing all options previously available only in system tray.
  * KV Cache configuration (Auto / Min / Medium / Max / Custom) both through the Settings page and system tray.
  * Improved model management controls: clearing model performance data and auto-removal behavior.
</Update>

<Update label="⎔ Capsule 0.2.7" description="2025-12-02" tags={["⎔ Capsule"]}>
  ### Added

  * Added environment-based configuration support.
  * Added support for multi-GPU cache size calculation.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.7" description="2025-12-01" tags={["።‖ Fortytwo P2P App"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.5" description="2025-11-27" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Added model list versioning. Current version is now visible on the About page.

  ### Changed

  * Locally stored encrypted private key file is now renamed and hidden.
  * Reduced frequency of some notifications. VPN alerts should no longer appear too often.

  ### Fixed

  * Corrected model size conversion (GiB → GB) for Hugging Face models.
  * Hidden unnecessary update messages (checking for updates, no updates) during normal runtime.
  * Restart button now functions correctly.
</Update>

<Update label="⎔ Capsule 0.2.6" description="2025-11-24" tags={["⎔ Capsule"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label="⏃ Protocol 0.18.0" description="2025-11-24" tags={["⏃ Protocol"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.4" description="2025-11-24" tags={["።‖ Fortytwo P2P App"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.11.24.0" description="2025-11-24" tags={[">_ Fortytwo P2P CLI"]}>
  ### Changed

  * Updated server endpoints.
</Update>

<Update label="⬢ Featured Models 1.0.0" description="2025-11-20" tags={["⬢ Featured Models"]}>
  ### Added

  * Added new featured model: **VibeThinker 1.5B**: Efficient reasoning performance with math and coding problems; suitable for nodes with tight memory.<br />
    *REPO = mradermacher/VibeThinker-1.5B-GGUF*<br />
    *MODEL NAME = VibeThinker-1.5B.Q4\_K\_M.gguf*
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.11.20.0" description="2025-11-20" tags={[">_ Fortytwo P2P CLI"]}>
  ### Added

  * Added new featured model: **VibeThinker 1.5B**: Efficient reasoning performance with math and coding problems; suitable for nodes with tight memory.
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.11.17.0" description="2025-11-17" tags={[">_ Fortytwo P2P CLI"]}>
  ### Changed

  * (Windows) Extended cleanup when updating the capsule.
</Update>

<Update label="⎔ Capsule 0.2.5" description="2025-11-14" tags={["⎔ Capsule"]}>
  ### Fixed

  * Fixed issues that caused the Capsule to stop without an exit code on Windows systems with Nvidia RTX 40xx and 50xx series.
</Update>

<Update label="⏃ Protocol 0.17.0" description="2025-11-05" tags={["⏃ Protocol"]}>
  ### Fixed

  * Fixed an issue that caused a loss of some requests in the network.
</Update>

<Update label="⎔ Capsule 0.2.4" description="2025-11-05" tags={["⎔ Capsule"]}>
  ### Fixed

  * Fixed an issue that caused a `Model produced no output` error to be displayed incorrectly.
</Update>

<Update label="⏃ Protocol 0.16.0" description="2025-11-05" tags={["⏃ Protocol"]}>
  ### Fixed

  * Network traffic and stability optimization.
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.10.28.0" description="2025-10-28" tags={[">_ Fortytwo P2P CLI"]}>
  *Improved readability, Windows layout fix.*

  ### Added

  * General readability improvements, style unification and extra spacings added.
  * (Windows) SETTINGS symbol added instead of a symbol that caused visualization issues.
  * Settings now have a header when activated.
  * Auto-Select mode now has a header when activated.

  ### Changed

  * "You chose" text has been changed to "Model selected" in the model selection option.
</Update>

<Update label="⏃ Protocol 0.15.3" description="2025-10-27" tags={["⏃ Protocol"]}>
  *Better logs readability.*

  ### Changed

  * "New request found" log entries are now summarized to not overload the CLI log.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.3" description="2025-10-23" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Added a new field to the Bug Report form, `Specify public wallet address`. Filling out this field can substantially speed up resolving an issue.

  ### Fixed

  * Architecture for Linux version now correctly shows x64 instead of ARM64.
  * Fixed an issue with activation code functionality during authorization.
  * Fixed an issue where “Insufficient MON” notification was displayed when a user had sufficient MON in their wallet.
</Update>

<Update label="⏃ Protocol 0.15.2" description="2025-10-20" tags={["⏃ Protocol"]}>
  *Better logs readability.*

  ### Changed

  * Improved logs readability by reducing the number of excessive log entries and removing unnecessary details.
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.10.17.0" description="2025-10-17" tags={[">_ Fortytwo P2P CLI"]}>
  *Strand-Rust-Coder-14B-v1 is now available in the CLI App. Our first Fortytwo-native model built with the community; state-of-the-art for Rust code generation.*

  You can find more details on the [model's Hugging Face page <Icon icon="arrow-up-right" />](https://huggingface.co/Fortytwo-Network/Strand-Rust-Coder-14B-v1-GGUF).

  ### Added

  * **Strand-Rust-Coder-14B-v1** is available under 21st position in the model selection list.

  ### Fixed

  * Fixed custom model import from Hugging Face for Windows.
</Update>

<Update label="⬢ Featured Models 0.1.0" description="2025-10-17" tags={["⬢ Featured Models"]}>
  ### Added

  * Added new rust model **Fortytwo\_Strand-Rust-Coder-14B-v1**: Built by Fortytwo: Rust specialist that outputs idiomatic, compile-ready code and handles fixes/refactors; state-of-the-art on Rust benchmarks.<br />
    *REPO = Fortytwo-Network/Strand-Rust-Coder-14B-v1-GGUF*<br />
    *MODEL NAME = Fortytwo\_Strand-Rust-Coder-14B-v1-Q4\_K\_M.gguf*

  ### Changed

  * (Fortytwo P2P CLI) Strand-Rust-Coder-14B-v1 now takes position number **21** instead of **Tessa-Rust-T1-7B**.

  ### Removed

  * **Tessa-Rust-T1-7B**<br />
    *bartowski/Tesslate\_Tessa-Rust-T1-7B-GGUF*<br />
    *Tesslate\_Tessa-Rust-T1-7B-Q4\_K\_M.gguf*
</Update>

<Update label=">_ CLI 2025.10.16.0" description="2025-10-16" tags={[">_ Fortytwo P2P CLI"]}>
  *Added support of features from the ⎔ Capsule 0.2.3 release.*

  ### Added

  * **KV Cache** can now be manually defined.
    * The option can be found under Settings category in the model selection menu.
    * The option must be defined each time the script is restarted.
    * Default option on start-up has `mode` set to `auto`, allocating as much available resources as possible for caching.
  * Added support for **local GGUF model import** under Import Custom option in the model selection menu list.

  ### Changed

  * **Settings** menu option added to the model selection list as a **\[0]** option.
  * **Import Custom** now provides options: **\[1] Hugging Face Import**  or **\[2] Local GGUF Model Import**.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.2" description="2025-10-17" tags={["።‖ Fortytwo P2P App"]}>
  *Strand-Rust-Coder-14B-v1 is now available in the App. Our first Fortytwo-native model built with the community; state-of-the-art for Rust code generation.*

  You can find more details on the [model's Hugging Face page <Icon icon="arrow-up-right" />](https://huggingface.co/Fortytwo-Network/Strand-Rust-Coder-14B-v1-GGUF).

  ### Added

  * macOS version is now signed and supports auto-updates.
  * New app update checking flow: startup update screen has been redesigned; the app now also downloads updates for protocol/capsule during the initial update.
  * New model card view — you can now set the model to immediately enable upon download completion.
  * Improved model manager performance — model list is now cached in the storage folder and updates automatically; models and updates should now be accessible in most countries without a VPN.
  * Model manager stabilization — improved reliability when pausing or cancelling downloads.

  ### Fixed

  * Fixed an issue causing Swapped Identity to appear for some users after an update.
  * Fixed duplicate “resolved” notifications appearing multiple times.
  * Fixed fullscreen notifications not showing on the Checking for Updates screen.
  * Fixed an issue where quitting the app sometimes took longer than expected.
  * Fixed menu bar app icon color for Linux version.

  ### macOS Users

  <Warning>Important Update Instructions</Warning>

  Version 0.2.2 is now code-signed, which means its internal app identity has changed.
  Because of this, the new version cannot access data previously stored in the Keychain by the unsigned v0.2.1 build.
  Fortytwo continues to use macOS Keychain to securely store your private key at system level.

  When launching the new version, macOS may ask for your system password to grant Fortytwo access to its secure data — you have two options:
  you can safely enter your password and click `Always Allow` to avoid future prompts, or you can delete the previous data and restore your profile in the new version.

  #### Steps for macOS users to reset the app data

  <Steps>
    <Step title="Export your private key">
      <Warning>Important! If you haven't already saved your recovery phrase or at least your private key, export your private key now — otherwise, you will lose access to your account and your node.</Warning>

      How to export your private key:

      1. Right click on the Fortytwo P2P App icon in the Menu Bar.
      2. Select `Account` > `Export Private Key`.
      3. Select destination and confirm.

      Make sure to save it to a secure location.
    </Step>

    <Step title="Delete the old record">
      Open the Terminal and run this command:

      ```bash Zsh theme={null}
      	security delete-generic-password -s "Fortytwo Safe Storage" || true
      ```
    </Step>

    <Step title="Install the latest version of the Fortytwo P2P App">
      [Download <Icon icon="download" />](https://downloads.fortytwo.network/updates/latest-mac) and install the latest signed macOS version (0.2.2 and above).
    </Step>

    <Step title="Launch the App">
      Run the app and sign in again using your private key or secret recovery phrase.
    </Step>
  </Steps>
</Update>

<Update label=">_ Fortytwo P2P CLI 2025.10.08.0" description="2025-10-08" tags={[">_ Fortytwo P2P CLI"]}>
  ### Added

  * Added connection validation.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.1" description="2025-10-06" tags={["።‖ Fortytwo P2P App"]}>
  ### Fixed

  * Fix an issue where best-fit model was detected incorrectly and caused errors and unintended behavior.
  * Fix cache cleanup when canceling a model download.
</Update>

<Update label="።‖ Fortytwo P2P App 0.2.0" description="2025-10-06" tags={["።‖ Fortytwo P2P App"]}>
  *Introducing manual model management mode in the Fortytwo P2P App. It is now possible to conveniently switch between auto and manual modes at any time.*

  ### Added

  * Auto/Manual mode switch with an onboarding stage explaining the modes.
  * Main screen now includes:
    * Model cards.
    * Controls to switch between the modes.
    * Model search.
  * Manual mode features:
    * Model search. Search for models from the Fortytwo P2P App featured list, Hugging Face, or your local imported models.
    * Model download and deletion.
    * Model activation — defining active model from the list of downloaded and imported ones.
    * Setting the model to activate when downloaded or imported.
    * Featured models can get tagged as Best-fit, Recommended, Slow, or Unfit to aid with the right model selection for the system.
    * Model cards list recommended VRAM/RAM sizes for their respective models.
    * Local import support for .GGUF model files.

  ### Fixed

  * Corrected free storage calculation in tray menu.
  * Added **Return to authorization options** button on the **Activation code** step — allows returning if the code is invalid.
</Update>

<Update label="⏃ Protocol 0.15.1" description="2025-10-03" tags={["⏃ Protocol"]}>
  ### Changed

  * Simplified and clarified the way the node reports round results in its metrics.
</Update>

<Update label="⎔ Capsule 0.2.3" description="2025-09-24" tags={["⎔ Capsule"]}>
  ### Added

  *The following has been added for the capsule to support upcoming app features.*

  * Local model and embeddings models loading to unlock faster cold starts:
    * `--llm-model-path <path>`
    * `--embeddings-model-path <path>`
  * KV-cache size controls to control how much of your system resources are allocated to inference generation. Default: `--kv-size-mode auto`
    * `--kv-size-mode <auto|min|medium|max>` (apply sizing by mode: auto | 33% | 66% | 100% of available limit)
    * `--kv-size-tokens <int>` (default and min are `1024`, target cache in tokens, fallback to default: `--kv-size-mode auto`)
    * `--kv-size-gb <float>` (default is `1.0`, target cache in GB. If less than `1`, fallback to `--kv-size-mode` or `--kv-size-mode auto`)
</Update>

<Update label="⏃ Protocol 0.15.0" description="2025-09-23" tags={["⏃ Protocol"]}>
  ### Fixed

  * Fixed intent resolution comparison.
</Update>

<Update label="⏃ Protocol 0.14.0" description="2025-09-16" tags={["⏃ Protocol"]}>
  ### Added

  * Faster, more resilient networking: QUIC transport (opt-in) runs alongside TCP on the same port with automatic TCP fallback; upgraded to QUICv1 for lower latency and smoother handshakes.
  * Connect through tough networks: NAT hole-punching (DCUtR) upgrades relayed links to direct connections with retry/backoff and Prometheus metrics for visibility.
  * Smarter peer knowledge: PeerStore now tracks multiple addresses per peer to improve reachability.
  * Configurable relay mode: Optional GossipSub relay acceptance/validation (off by default). Enable with `FT_GOSSIPSUB_RELAY_ENABLED=true`.
  * Easier bring-up & testing: IPFS bootstrapping option; Docker Compose to run multiple nodes on one machine; Dockerfile now works with Podman.
  * Better observability: New Prometheus metrics for DCUtR operations, peer analysis, and connection state.

  ### Changed

  * Less waste, better throughput: Participation logic tuned so a node skips very slow or ineffective rounds.
  * Lower latency by default: Dialer prioritizes QUIC addresses when available.
  * Cleaner logs: Error noise reduced where retries are automatic; clearer transport-type logging.
  * Smarter routing: Improved address policy for relayed vs. direct connections.
  * Safer defaults: Updated node/network configuration; DCUtR is enabled conditionally via config.
</Update>

<Update label="።‖ Fortytwo P2P App 0.1.2" description="2025-09-16" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Extra validation before capsule auto-start so it doesn't interfere with app authorization process.
  * Compatibility check: if nvidia-smi is missing and the system is not macOS, the app now shows System Incompatible notification.

  ### Changed

  * GPU-based systems will now always run GPU capsule.
  * Simplified onboarding: `Go to node` button now closes the dialog instead of redirecting to another page.
  * (macOS) Improved active window behavior:
    * App window always creates a menu bar entry.
    * When switching to app, a menu bar window is opened.
    * When launching the app when it is already running, its window is opened instead of launching another instance of the app.

  ### Fixed

  * Notifications:
    * Removed interaction options from notifications that have already been resolved.
    * Fixed globe icon color on notifications — it was invisible in light theme.
</Update>

<Update label="።‖ Fortytwo P2P App 0.1.1" description="2025-09-10" tags={["።‖ Fortytwo P2P App"]}>
  ### Added

  * Bug reporting: System information is now collected and sent as a separate file (contained in the log archive) when submitting a bug report.
  * Added "Documentation" item to the tray menu.
  * (macOS) Launch Agent is now used instead of AppleScript for the auto-launch option — now the app should't ask for additional permissions.

  ### Changed

  * (macOS) Updated app icons.
  * Clarified titles in a few app messages.
  * Adjusted memory check — it now uses available VRAM instead of total VRAM.
  * VRAM warning is now triggered only if less than 2 GB of free VRAM is available.

  ### Fixed

  * Notification Center:
    * Some feedback messages no longer appear in the notification center for clarity.
    * Some warnings and critical notifications (e.g. Heavy Load, Connection Failed) will no longer duplicate if an active notification with the same code already exists.
  * Fixed Connection Failed notification during onboarding – now only shown at the creation stage when an internet connection is actually needed.
  * Fixed model list handling — now filtering out unnecessary models using the `automation:false` tag, they will no longer appear in automatic mode.
  * Improved stability when losing internet connection:
    * When a download is interrupted due to lost connection, the model will no longer be mistakenly marked as downloaded and re-downloaded by the capsule.
    * Better separation of Connection Failed error states during node operation.
    * Reconnection notifications are now more stable both when the capsule is stopped and when it is running. It retries pings and disappears when connection is established; if automatically paused, it resumes automatically.
    * Service connection status no longer disappears after a timeout when offline — it now clears only after a successful ping or when the node connection issue is resolved.
    * Downloads now reliably restart once internet connection is restored (previously, this worked inconsistently).
  * Fixed broken dock menu on macOS
  * Fixed incorrect behavior of tray options in "Open in window" mode.
</Update>

<Update label="⏃ Protocol 0.13.0" description="2025-09-09" tags={["⏃ Protocol"]}>
  ### Changed

  * Increased maximum round capacity
</Update>

<Update label="⎔ Capsule 0.2.2" description="2025-09-09" tags={["⎔ Capsule"]}>
  ### Added

  * Added machine specs logs.
  * Added kv-cache init log.

  ### Changed

  * Fixed gguf file merging.
</Update>

<Update label="⏃ Protocol 0.12.1" description="2025-09-01" tags={["⏃ Protocol"]}>
  ### Fixed

  * Price calculation when creating a request through rpc API has been fixed.
</Update>

<Update label="⏃ Protocol 0.12.0" description="2025-08-20" tags={["⏃ Protocol"]}>
  ### Added

  * Add `--announce` flag.

  ### Changed

  * Removed propagation to swarm and inference manager during createIntent.

  ### Fixed

  * Creating an intent on behalf of another author via RPC API.
</Update>

<Update label="⎔ Capsule 0.2.1" description="2025-08-20" tags={["⎔ Capsule"]}>
  ### Added

  * Added support gpt-oss.
  * Updated metadata values for symbols.

  ### Changed

  * Updated llama cpp bindings to b6118.
</Update>

<Update label="⏃ Protocol 0.11.0" description="2025-08-14" tags={["⏃ Protocol"]}>
  ### Added

  * Implemented new ranking algorithm.
  * Added warning message when node TPS is below optimal.

  ### Fixed

  * Fixed gossip performance metrics.
</Update>

<Update label="⏃ Protocol 0.10.0" description="2025-08-05" tags={["⏃ Protocol"]}>
  ### Added

  * Startup timeout (180s) to prevent hangs during initialization.
  * Detailed startup logs for Swarm and Blockchain readiness.
  * New exit codes:
    * 10: Startup error
    * 11: Startup timeout
  * Added delegator address to rpc API.

  ### Changed

  * Extended capsule API for making participation decisions.

  ### Fixed

  * Readiness check loop now handles shutdowns correctly, avoiding hot loops and high CPU load.
</Update>

<Update label="⎔ Capsule 0.2.0" description="2025-08-04" tags={["⎔ Capsule"]}>
  ### Added

  * Added health check to API.
  * Added non-blocking metadata request.
  * Added context length checker to evaluate-participation request.
  * Added support structured outputs.
  * Added new ranking schema for v1 ranking.
  * Implemented Efficient KV Cache Reuse for ranking.
</Update>

<Update label="⎔ Capsule 0.1.34" description="2025-07-16" tags={["⎔ Capsule"]}>
  *Improved offline support and startup observability for capsule workers.*

  ### Added

  * Offline mode support for using locally cached models.
  * Notification system to confirm successful worker initialization before warm-up.

  ### Changed

  * Adjusted worker initialization sequence for improved launch stability.
</Update>

<Update label="⏃ Protocol 0.9.0" description="2025-07-16" tags={["⏃ Protocol"]}>
  *Improved chain sync robustness and streamlined network module logic.*

  ### Added

  * Async block gap processing to detect and fill missing blockchain data automatically.
  * Minor stability improvements for consensus submissions.

  ### Changed

  * Removed relay-specific logic from the network module to simplify architecture.
</Update>

<Update label="⏃ Protocol 0.8.4" description="2025-07-10" tags={["⏃ Protocol"]}>
  *Corrected label assignments in the inference metrics output.*

  ### Fixed

  * Corrected label assignments in the inference metrics output.
</Update>

<Update label="⏃ Protocol 0.8.3" description="2025-07-09" tags={["⏃ Protocol"]}>
  *Enhanced observability, RPC efficiency, and improved compatibility for CLI/GUI workflows.*

  ### Added

  * Protocol exit codes for improved error signaling.
  * Optimized RPC by implementing raw block polling with bloom filter pre-checks to reduce event overhead.
  * Added operational metrics (node state, request tracking).
  * Included blockchain-specific metrics.

  ### Changed

  * Updated console time format for improved readability.
  * Reintroduced CLI arguments for backward compatibility with CLI and GUI apps.
  * Upgraded Rust version to 1.85.0.

  ### Fixed

  * Corrected RPC request handling to ensure proper use of swarm consensus.
</Update>

<Update label="⎔ Capsule 0.1.33" description="2025-07-03" tags={["⎔ Capsule"]}>
  *Improved process management with graceful shutdown and defined exit codes.*

  ### Added

  * Graceful shutdown handling for clean termination.
  * Defined exit codes for clearer process state signaling and bug reporting.
</Update>

<Update label="⎔ Capsule 0.1.32" description="2025-07-02" tags={["⎔ Capsule"]}>
  *Introduced web crawling support and updated Python worker behavior.*

  ### Added

  * Web crawler integration.

  ### Changed

  * Disabled Python worker by default.
</Update>

<Update label="⏃ Protocol 0.8.2" description="2025-07-01" tags={["⏃ Protocol"]}>
  *Introduced robust failover, improved ranking, and overall improvements to networking stability*

  ### Added

  * Implemented polling-based staggered submission for inference resolutions with position-based delays.
  * Added initial version of new ranking algorithm.
  * Added fallback RPC support with automatic failover and retry layer for improved blockchain connectivity.

  ### Changed

  * Changed blockchain event query error logging from error to warning level since query\_events is automatically retried.
  * Changed swarm connection loss logging from error to warning level when insufficient peers are available.
  * Changed capsule binding address from localhost to 0.0.0.0 to allow access from external clients.
</Update>

<Update label="⎔ Capsule 0.1.315" description="2025-06-2" tags={["⎔ Capsule"]}>
  *Dependency upgrade for improved compatibility and performance.*

  ### Changed

  * Updated core dependencies: llama.cpp and ft-tools.
  * Upgraded Rust version to 1.85.0.
</Update>

<Update label="⏃ Protocol 0.8.1" description="2025-06-24" tags={["⏃ Protocol"]}>
  *Improved transaction reliability and deadline handling through buffer tuning and multi-bootstrap support.*

  ### Added

  * Introduced support for multiple bootstrap nodes to improve peer connectivity.

  ### Changed

  * Increased blockchain safety buffer and inference submission offset for better transaction stability.
</Update>

<Update label="⏃ Protocol 0.8.0" description="2025-06-12" tags={["⏃ Protocol"]}>
  *Major networking upgrade with peer discovery, sync improvements, and cleanup of blockchain and log systems.*

  ### Added

  * Introduced /metrics endpoint for Prometheus-based performance analysis.
  * Enabled new peer discovery method during node initialization.
  * Added node startup synchronization.
  * Integrated Identify protocol for peer recognition.
  * Forwarded sorted ranking results in HTTP responses.

  ### Changed

  * Renamed 42T token to FOR.
  * Updated bootstrap nodes.
  * Upgraded Rust toolchain to 1.84.0.

  ### Fixed

  * Reworked network interaction to resolve sync issues.
  * Ignored early requests received before node startup completed.
  * Cleaned up log to reduce noise.
</Update>

<Update label="⏃ Protocol 0.6.0" description="2025-05-22" tags={["⏃ Protocol"]}>
  *Introduced new peer discovery method and startup synchronization with major networking and blockchain module updates.*

  ### Added

  * Integrated Kademlia protocol for peer discovery during initialization.
  * Implemented node startup synchronization to ensure readiness.
  * Added Identify protocol support for recognizing peer nodes.
  * Forwarded sorted ranking results included in HTTP response.

  ### Changed

  * Refactored blockchain module to use alloy instead of ethers.

  ### Fixed

  * Resolved bootstrap node initialization issue.
  * Ignored incoming requests created before node startup is complete.
</Update>

<Update label="⎔ Capsule 0.1.27" description="2025-05-20" tags={["⎔ Capsule"]}>
  *Fixed warm-up logic to improve LLM readiness at startup.*

  ### Fixed

  * Resolved issues with LLM warm-up during capsule initialization.
</Update>

<Update label="⎔ Capsule 0.1.26" description="2025-05-14" tags={["⎔ Capsule"]}>
  *Improved development observability.*

  ### Changed

  * Added additional logs for development and debugging purposes.
</Update>

<Update label="⎔ Capsule 0.1.25" description="2025-05-12" tags={["⎔ Capsule"]}>
  *Switched default tokenizer configuration.*

  ### Changed

  * Changed default tokenizer used for completions.
</Update>

<Update label="⏃ Protocol 0.5.4" description="2025-05-12" tags={["⏃ Protocol"]}>
  *Added request sizing parameter to capsule completions.*

  ### Added

  * Included max\_size parameter in the completions request to the capsule for better input control.
</Update>

<Update label="⎔ Capsule 0.1.24" description="2025-05-09" tags={["⎔ Capsule"]}>
  *Added support for reasoning token controls and max output size.*

  ### Added

  * Option to include reasoning tokens in output.
  * Parameter to adjust reasoning token count.
  * max\_size field added to completion requests.
</Update>

<Update label="⎔ Capsule 0.1.23" description="2025-05-07" tags={["⎔ Capsule"]}>
  *Increased reserved memory to improve runtime reliability.*

  ### Changed

  * Increased default reserved memory allocation.
</Update>

<Update label="⏃ Protocol 0.5.3" description="2025-05-07" tags={["⏃ Protocol"]}>
  *Added request size cap to prevent overload in devs request.*

  ### Added

  * max\_size field to the /devs request for controlled query sizing.
</Update>

<Update label="⎔ Capsule 0.1.22" description="2025-05-07" tags={["⎔ Capsule"]}>
  *Major performance and capability upgrade including dynamic context, GPU offloading, and embeddings truncation.*

  ### Added

  * Embedding truncation for improved memory efficiency.
  * Support for using the model's maximum context size.
  * Optimized input token decoding for better inference speed.
  * Tokenizer object now returned in /metadata.
  * Dynamic selection mechanism for context length.
  * GPU offloading support for KV cache.

  ### Changed

  * Updated development dependencies.

  ### Fixed

  * Fixed batch prefill logic to improve prompt handling stability.
</Update>

<Update label="⏃ Protocol 0.5.2" description="2025-05-01" tags={["⏃ Protocol"]}>
  *Made tokenizer ID optional in developer-facing requests.*

  ### Fixed

  * Tokenizer\_id in /devs request is now optional, improving support for lightweight queries.
</Update>

<Update label="⏃ Protocol 0.5.1" description="2025-05-01" tags={["⏃ Protocol"]}>
  *Improved handling for empty tokenizer values in intents.*

  ### Fixed

  * Correctly handles empty tokenizer cases in submitted intents.
</Update>

<Update label="⏃ Protocol 0.5.0" description="2025-04-30" tags={["⏃ Protocol"]}>
  *Major feature release with metadata access, performance estimation, new ranking strategies, and node load balancing.*

  ### Added

  * GET /metadata endpoint for capsules to fetch version, token limits, and tokenizer info.
  * Participation now estimates performance limits from capsule metadata.
  * New ranking method added that returns a confidence score per candidate.
  * Load balancing system introduced to evenly distribute requests across nodes.
  * New experimental pairwise ranking algorithm implemented.
  * Failsafe added to exit if allowance approval fails at minimum threshold.
</Update>

<Update label="⎔ Capsule 0.1.21" description="2025-04-30" tags={["⎔ Capsule"]}>
  *Added support for Qwen3 ARC model and updated dev dependencies.*

  ### Added

  * Added Qwen3 model support.

  ### Changed

  * Updated development dependencies.
</Update>

<Update label="⎔ Capsule 0.1.20" description="2025-04-11" tags={["⎔ Capsule"]}>
  *Improved system prompt and added error handling for model outputs.*

  ### Changed

  * Added handling for empty model output in get\_completions response.
  * Updated system prompt for better context initialization.
</Update>

<Update label="⎔ Capsule 0.1.19" description="2025-04-11" tags={["⎔ Capsule"]}>
  *Disabled Python execution and introduced structured inputs for participation logic.*

  ### Changed

  * Temporarily disabled Python interpreter.
  * Fixed emoji rendering issue in output.
  * Added structured input format for evaluate-participation.
</Update>

<Update label="⏃ Protocol 0.4.0" description="2025-04-10" tags={["⏃ Protocol"]}>
  *Improved token handling and pre-checks before participation.*

  ### Added

  * Node balance validation before joining a round.
  * Auto-approval of 42T token allowance for smoother participation.
</Update>

<Update label="⎔ Capsule 0.1.18" description="2025-04-10" tags={["⎔ Capsule"]}>
  *Introduced new ranking endpoint and support for structured input.*

  ### Added

  * New /v2/ranking API endpoint.
  * Support for structured input objects.

  ### Changed

  * Ranking results are now derived directly from logits.
</Update>

<Update label="⎔ Capsule 0.1.17" description="2025-04-09" tags={["⎔ Capsule"]}>
  *Introduced a new metadata endpoint and transitioned to mpsc-based streaming.*

  ### Added

  * /metadata endpoint for diagnostics and external tooling.

  ### Changed

  * Replaced Notify/result store with mpsc streaming for improved data flow.
</Update>

<Update label="⎔ Capsule 0.1.16" description="2025-04-08" tags={["⎔ Capsule"]}>
  *Extended capabilities with web search, Python execution.*

  ### Added

  * Brave API integration for web search.
  * Python execution support.

  ### Changed

  * Updated Rust version to 1.84.0.
</Update>

<Update label="⏃ Protocol 0.3.2" description="2025-04-07" tags={["⏃ Protocol"]}>
  *Adjusted participation logic and timing for better request flow.*

  ### Changed

  * Increased max duration for inference join attempts.
  * Modified participation probability to retain a proportional number of idle nodes for upcoming requests.
</Update>

<Update label="⏃ Protocol 0.3.1" description="2025-04-04" tags={["⏃ Protocol"]}>
  *Improved gas efficiency and fixed retry logic.*

  ### Changed

  * Optimized client-side gas consumption.

  ### Fixed

  * Patched retry logic covering blockchain context, balances, token approvals, and intent resolution.
</Update>

<Update label="⎔ Capsule 0.1.15" description="2025-04-04" tags={["⎔ Capsule"]}>
  *Performance improvements, private crate support, and prompt tuning.*

  ### Added

  * Support for custom private Rust crates.
  * Token generation speed: 97.63% parity with llama.cpp.
  * Performance metrics logged in debug output.

  ### Changed

  * Codebase refactor for modularity and readability.
  * Updated ranking prompt and generation parameters.
  * Updated papaya crate to v0.2.1.
  * Separated token-to-text from generation worker.
</Update>

<Update label="⏃ Protocol 0.3.0" description="2025-04-02" tags={["⏃ Protocol"]}>
  *Introduced off-chain consensus and automated infrastructure upgrades.*

  ### Added

  * Off-chain consensus layer via P2P.
  * Automated deployment pipeline for new releases.

  ### Changed

  * Refactored message queue to ignore certain request messages and prevent storage overhead.
</Update>

<Update label="⎔ Capsule 0.1.14" description="2025-03-24" tags={["⎔ Capsule"]}>
  *Internal refactor with focus on stability and efficiency.*

  ### Changed

  * Code refactor for maintainability.

  ### Fixed

  * Runtime resource usage.
</Update>

<Update label="⎔ Capsule 0.1.13" description="2025-03-20" tags={["⎔ Capsule"]}>
  *Feature update with a new embeddings model, flash attention and improved ranking capabilities.*

  ### Added

  * New embeddings model.
  * Flash attention for faster processing.
  * max\_tokens support for ranking requests.

  ### Changed

  * Expanded context window for embeddings.
    ranking logic.

  ### Fixed

  * LLM context window bug.
  * General resource optimization.
</Update>

<Update label="⎔ Capsule 0.1.12" description="2025-03-19" tags={["⎔ Capsule"]}>
  *Enabled model upload cache configuration*

  ### Added

  * Configurable cache location for uploaded models.
</Update>

<Update label="⎔ Capsule 0.1.11" description="2025-03-17" tags={["⎔ Capsule"]}>
  *Backend upgrades and CUDA environment improvements.*

  ### Changed

  * Upgraded to llama.cpp version b4902.
  * Updated Linux CUDA configuration.
</Update>

<Update label="⏃ Protocol 0.2.2" description="2025-03-13" tags={["⏃ Protocol"]}>
  *Temporary gas optimizations and threading fix.*

  ### Changed

  * Reduced MAX\_PARTICIPATION\_AMOUNT from 12 to 7 to lower gas costs.

  ### Fixed

  * Fixed multithreading desync issues in active round processing.
</Update>

<Update label="⏃ Protocol 0.2.1" description="2025-03-12" tags={["⏃ Protocol"]}>
  *Minor runtime adjustment to stabilize startup.*

  ### Added

  * Added delay before initializing BlockchainBridge.
</Update>

<Update label="⏃ Protocol 0.2.0" description="2025-03-12" tags={["⏃ Protocol"]}>
  *Major enhancements to blockchain integration, formatting, and dependencies.*

  ### Added

  * Retry logic for blockchain sync, balances, approvals, and intents via retry\_on\_error!.
  * New u256\_frac\_mul! macro for precise U256 math.
  * Time sync with on-chain deadlines for accurate state validation.

  ### Changed

  * Code formatting improvements via rustfmt.toml.
  * Dependency Updates
  * Upgraded rand crate from 0.8 → 0.9.0.
  * Added semver crate at 1.0.0.
</Update>

<Update label="⏃ Protocol 0.1.33" description="2025-03-11" tags={["⏃ Protocol"]}>
  *Ranking module hotfix*

  ### Fixed

  * Optimized mathematical functions in the ranking module.
</Update>

<Update label="⏃ Protocol 0.1.32" description="2025-03-10" tags={["⏃ Protocol"]}>
  *Improved swarm stability, fixed reconnection and ranking issues, and enhanced logging.*

  ### Added

  * Implemented forced exit on swarm disconnection to improve network stability.

  ### Changed

  * Enhanced log messages for better debugging and clarity.

  ### Fixed

  * Resolved an issue causing failures in swarm reconnection.
  * Corrected the ranking algorithm for improved inference accuracy.
  * Fixed a bug where ParticipateInInference continued running unexpectedly.
</Update>