Inferencer - Private AI Studio
Advanced Local AI Assistant
Free · In‑App Purchases
Inferencer lets you run, host and deeply control the latest SOTA AI models (DeepSeek, GLM, Kimi, MiniMax, Nemotron, Qwen and more) on your own computer.
No data is sent to the cloud for processing - maintaining your complete privacy.
Advanced inferencing controls give you complete control on accuracy and outputs.
Models
Start in the models section where you can select the location of existing models or download new ones directly from Hugging Face or ModelScope.
Use the distributed compute feature to load a model across two or more Macs, or use the model streaming feature to inference larger models partially from storage.
Chats
Select a model to inference and enter a prompt to begin. At any point you can switch between models and continue the chat to see what else they can uncover. You can also selectively delete past messages to keep the model focused.
Attachments
Drag and drop in file attachments such as PDF, Word, TXT, Code, Images and Audio. Multimodal models can inference the images and audio directly, or alternatively automatic OCR and feature extraction will be performed for Text only inference support.
Inference Controls
Control the inferencing parameters including batching and multiprocessing to inference multiple generations and models at the same time, thinking limits, intensity of processing, and context quantization to further reduce the memory usage and model streaming to load models larger than available memory.
Token Entropy, Inspection and Control
Select the inspectors to peek into the inner-workings of each word outputted and see the model's confidence levels and alternative choices. Utilise the response control feature to direct the output the model generates. For example, skipping the preamble or directing the model to output in structured html.
Tools
The tools editor allows you to enable built in tools such as get_webpage_content or add in your own, so that models can use them when needed. For example, if you'd like a webpage or search result inferenced, simply enable the tool in the Tools section, and allow tool calls in the chat settings panel.
Server
If enabled, the server feature allows you to serve and connect to your own or trusted devices. No data is sent elsewhere. Also includes Compatible APIs for application development.
Prompt Caching
Significantly improves prompt processing by prefix-matching previously processed prompts. Uses a user-configurable cache pool with optional external storage and automatic LRU (least recently used) eviction (enabled in Settings).
Distributed Inference
With distributed compute you can link together multiple Macs, sharing the memory to inference larger models. To use, make sure it's enabled in each of the Server's Serve settings. Connect directly or setup a multi-node Cluster. Once a connection is made, if the computers have the same model, a distributed compute icon will appear. Simply tap on it to load the model for distributed compute.
Coding Tools
Built-in support for Xcode Intelligence and Visual Studio Code. Use the server feature with Compatibility APIs enabled and SSL disabled to allow Xcode or Visual Studio Code to use Inferencer as a service provider.
Shortcuts
Use the Shortcuts app to automate inferencing workflows (e.g., copy text from clipboard > inference > speak result).
Settings
Includes parental controls, an automatic deletion policy and more.
Privacy
For maximum privacy, all AI processing happens offline and on your device, by default.
Subscriptions
Basic (Free): Most features unlocked for free including unlimited chats.
Professional: Upgrade for more advanced features including token inspection, distributed compute and model streaming.
Terms & Support
Terms of Use: inferencer.com/terms
Privacy Policy: inferencer.com/privacy
Disclaimer
Inferenced models may not always be accurate or contextually appropriate. You are responsible for verifying the information before making important decisions.
more App would not download models but the developer responded to email with a quick fix, and it should be fixed in next version. This is an interesting program - I haven’t figured out all the bells and whistles.
Developer Response Hi, an update was submitted to Apple earlier in the week to address an issue with downloading on macOS 26. As a workaround, you can download the models directly from HuggingFace and place them directly in your chosen Downloads folder. Feel free to report any other issues or requests you have on the public roadmap.
App would not download models but the developer responded to email with a quick fix, and it should be fixed in next version. This is an interesting program - I haven’t figured out all the bells and whistles.
Hi, an update was submitted to Apple earlier in the week to address an issue with downloading on macOS 26. As a workaround, you can download the models directly from HuggingFace and place them directly in your chosen Downloads folder. Feel free to report any other issues or requests you have on the public roadmap.
This app does something nothing else does: allow you to see alternative tokens in nearly real-time. And to adjust the output to your framing without complicated setup or per-model sidebar configuration. You can do it per conversation. For AI researchers, the curious, and those looking to tweak model outputs? This is a fabulous tool. Highly recommended for the curious and local LLM enthusiasts.
Developer Response Thank you so much for the support and feedback, if there's any other features you'd like implemented feel free to add it to the public roadmap.
This app does something nothing else does: allow you to see alternative tokens in nearly real-time. And to adjust the output to your framing without complicated setup or per-model sidebar configuration. You can do it per conversation. For AI researchers, the curious, and those looking to tweak model outputs? This is a fabulous tool. Highly recommended for the curious and local LLM enthusiasts.
Thank you so much for the support and feedback, if there's any other features you'd like implemented feel free to add it to the public roadmap.
As someone who has multiple Macs, I was looking forward to using this for distributed inference, but I think that charging a monthly subscription is a really bad business model. I would love to support the development of this program, but monthly fees are the exact kind of thing that I wanted to avoid by investing in local AI hardware. I hope you consider changing this to a one time payment.
Developer Response Hi Jack, thanks so much for sharing your perspective.
As someone who has multiple Macs, I was looking forward to using this for distributed inference, but I think that charging a monthly subscription is a really bad business model. I would love to support the development of this program, but monthly fees are the exact kind of thing that I wanted to avoid by investing in local AI hardware. I hope you consider changing this to a one time payment.
Hi Jack, thanks so much for sharing your perspective.
The maker of this app, xCreate is well beyond knowledgeable when it comes to inferencing LLM’s. Though some functions of using it as an agentic coding api endpoint needs work, the creator of this app is updating and fixing things on a literal daily basis. Keep in mind that he is allowing all of this to be used for free.
Developer Response Thanks so much for the support. If there's any specific issues you have with the api endpoints, please post them on the issues page or send over an email and we'll get it fixed for you.
The maker of this app, xCreate is well beyond knowledgeable when it comes to inferencing LLM’s. Though some functions of using it as an agentic coding api endpoint needs work, the creator of this app is updating and fixing things on a literal daily basis. Keep in mind that he is allowing all of this to be used for free.
Thanks so much for the support. If there's any specific issues you have with the api endpoints, please post them on the issues page or send over an email and we'll get it fixed for you.
+ DeepSeek V4.1 cluster support
+ Improvements to multi-node cluster caching
+ GLM 5.3 Flash multimodal cache support
+ Bug fixes and performance improvements
Also, in case you missed the recent updates:
+ Multi-node cluster support for Qwen vision and MiMo V2.6 vision
+ Answer Now button to dynamically end thinking
+ Inference multiple servers at the same time
+ Anthropic API and Claude Code support, including cache optimisations
+ MTP Thinking limit and multiprocessing support
+ Missed tool call detection (enabled by default)
+ HTML/JS validator and Python code viewer
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
2.4.4 20h ago
+ Multi-node cluster fixes and support for Qwen vision and MiMo V2.6 vision
+ Generation fix for multi-image inferences with caching enabled
+ WebGL Errors reporting and relative path relative path support for HTML viewer
+ Multiprocessing support for MTP
+ Bug fixes and performance improvements
Also, in case you missed the recent updates:
+ Support for DeepSeek V4.1, GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash, Step5, MiMo V2.6
+ Answer Now button to dynamically end thinking
+ Inference multiple servers at the same time
+ Anthropic API and Claude Code support, including cache optimisations
+ MTP Thinking limit support
+ Missed tool call detection (enabled by default)
+ HTML/JS validator and Python code viewer
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
2.4.3 6d ago
+ Multi-node cluster fixes and support for Qwen vision and MiMo V2.6 vision
+ Relative path support for HTML viewer
+ Multiprocessing support for MTP
+ Bug fixes and performance improvements
Also, in case you missed the recent updates:
+ Support for DeepSeek V4.1, GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash, Step5, MiMo V2.6
+ Answer Now button to dynamically end thinking
+ Inference multiple servers at the same time
+ Anthropic API and Claude Code support, including cache optimisations
+ MTP Thinking limit support
+ Missed tool call detection (enabled by default)
+ HTML/JS validator and Python code viewer
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
2.4.2 Oct 2
+ Answer Now button to dynamically end thinking
+ Support for Step5 Preview and MiMo V2.6
+ Inference multiple servers at the same time
+ Server switching integrated into models toolbar
+ MTP support for thinking limits
+ Python code viewer
+ Improvements to in-app tool calls
+ Tool calls made client side when inferenced via Server
+ Improvements to missed tool call detection and handling
+ Improvements to Markdown code blocks
+ Batching and multiprocessing hardening
+ Caching improvements for Qwen 3.8 and similar models
+ HTML import error improvements
+ Bug fixes and performance improvements
Also, in case you missed the recent updates:
+ Anthropic API and Claude Code support, including cache optimisations
+ OpenCode support for Server API date override
+ Missed tool call detection (enabled by default)
+ HTML/JS validator tool
+ File tool improvements
+ Support for DeepSeek V4.1, GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
2.4.1 Sep 27
+ Anthropic API and Claude Code support, including cache optimisations
+ Thinking limit improvements and enforcer option (disabled by default)
+ Thinking limits for Server API
+ OpenCode support for Server API date override
+ Missed tool call detection (enabled by default)
+ HTML/JS validator tool
+ File tool improvements
+ Cache re-use improvements
+ Vision cache support and tool call improvements for DeepSeek V4.1
+ Bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for DeepSeek V4.1, GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
2.4.0 Sep 17
+ Vision and Tool support for DeepSeek V4.1 (updated chat_template available)
+ Thinking improvements
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash, Hy4 and Ling-3.0-Flash
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Tool call streaming support
+ Improved support for importing Word docs
+ Rendering performance improvements for long generations
+ File tool fixes for Grep and find and replace
2.3.9 Sep 12
+ Initial support for DeepSeek V4.1
+ Thinking limit improvements
+ minMax and midMax thinking modes (max thinking mode with less tokens)
+ Improved support for importing Word docs
+ More rendering performance improvements for long generations
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash, Hy4 and Ling-3.0-Flash
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Tool call streaming support
+ File tool fixes for Grep and find and replace
2.3.8 Sep 11
+ Support for DeepSeek V4 Vision and Hy4
+ Thinking limit improvements and support for GLM 5.3
+ minMax thinking mode (max thinking mode with minimum tokens)
+ More rendering performance improvements for long generations
+ Tool call streaming support
+ File tool fixes for Grep and find and replace
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash and Ling-3.0-Flash
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
2.3.7 Sep 4
+ Support for GLM-5.3, GLM-5.3-Flash, Qwen3.8-Flash and Ling-3.0-Flash
+ Fix for DeepSeek MTP decoding
+ Thinking modes for GLM-5.3
+ Rendering performance improvements for long generations
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
2.3.6 Aug 30
+ Support for GLM-5.3-Flash, Qwen3.8-Flash and Ling-3.0-Flash
+ Improvements to thinking limits and end detector
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
2.3.5 Aug 30
+ Support for Qwen3.8-Flash and Ling-3.0-Flash
+ More Kimi K3 thinking modes
+ Improvements to End Generation detector
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
2.3.4 Aug 27
+ Qwen 3.8 xHigh inference improvements
+ Looping improvements for Qwen 3.8 and various models
+ Thinking detector improvements and thinking limit support for MTP
+ Caching improvements for Qwen 3.8 and various models
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for Kimi K3, DeepSeek V4 DSpark, Muse Glimmer and Ling-3.0-flash support
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
2.3.3 Aug 19
+ Qwen 3.8 xHigh, Muse Glimmer and Ling-3.0-flash support
+ More loop detection improvements
+ Kimi K3 Inference improvements
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for Kimi K3, DeepSeek V4 DSpark, Laguna S2.1 and Inkling
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
2.3.2 Aug 16
+ 17% Performance boost to DeepSeek DSpark
+ Fix for tool calls
+ More thinking detection improvements (enable to stop thinking loops)
+ Kimi K3 Inference improvements
+ Inference timeout and context limit settings
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for Kimi K3, DeepSeek V4 DSpark, Laguna S2.1 and Inkling
+ K3 Multimodal model streaming from storage
+ Improved Long Context Attention for DeepSeek and GLM
+ Server API Tool Call omission (disable Raw tokens in Server Settings)
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Pro Yearly plan (2 months free)
2.3.1 Aug 6
+ Kimi K3 and DeepSeek V4 DSpark support
+ K3 Multimodal model streaming from storage
+ Improved Context Attention for DeepSeek and GLM
+ Thinking Detection improvements (enable in Controls to stop thinking loops)
+ Server API Tool Call omission (disable Raw tokens in Server Settings)
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for Laguna S2.1, Inkling, Gemma 4 12B and MiniMax M3 Tool Calls
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ GLM 5.2 Long Context Improvements (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ Search in Conversation and
+ Copy and paste in images
+ Special tokens highlighted in Entropy view
+ File Attachments including support for Code, PDF, Word, and ePub
+ Text-only inference of images using metadata, OCR, and feature detection
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Pro Yearly plan (2 months free)
2.3.0 Aug 4
+ Thinking Detection (enable in Controls to stop thinking loops)
+ Copy and paste in images
+ Server API cache reuse optimisations for OpenClaw (enabled by default)
+ Server API context used now reports full context window
+ Excess prompt prefix fixes for Hy3 and other models
+ Support for Laguna S2.1, Inkling, Gemma 4 12B and MiniMax M3 Tool Calls
+ Qwen 3.6 27B tool call support
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ GLM 5.2 Long Context Improvements (enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ Search in Conversation
+ Special tokens highlighted in Entropy view
+ New chats per window/tab
+ Rename chats (available in right click context menu)
+ Downloading improvements including concurrency
+ File Attachments including support for Code, PDF, Word, and ePub
+ Text-only inference of images using metadata, OCR, and feature detection
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Pro Yearly plan (2 months free)
2.2.2 Jul 27
+ Support for Inkling, Gemma 4 12B and MiniMax M3 Tool Calls
+ GLM 5.2 Long Context Improvements (Extended EOG enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ Search in Conversation
+ Special tokens highlighted in Entropy view
+ New chats per window/tab
+ Rename chats (available in right click context menu)
+ Downloading improvements including concurrency
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ File Attachments including support for Code, PDF, Word, and ePub
+ Text-only inference of images using metadata, OCR, and feature detection
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Persistent prompt caching (enable in Settings - up to ~99x for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Pro Yearly plan (2 months free)
2.2.1 Jul 23
+ GLM 5.2 Long Context Improvements (Extended EOG enabled by default)
+ Thinking limits to avoid loops and reduce endless thinking gracefully
+ Search in Conversation
+ Special tokens highlighted in Entropy view
+ New chats per window/tab
+ Rename chats (available in right click context menu)
+ Downloading improvements including concurrency
+ Support for Gemma 4 12B and MiniMax M3 Tool Calls
+ Cache improvements for DeepSeek V4, Qwen 27B and more
+ Thinking detection for more models including Hy3, MiniMax M3, Ornith, and Step
+ Improved support for OpenCode
+ Support for Ollama api /ps
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ File Attachments including support for Code, PDF, Word, and ePub
+ Text-only inference of images using metadata, OCR, and feature detection
+ File Tools including Grep, Find and Replace, and Read/Write Operations
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Persistent prompt caching (enable in Settings - up to ~99x for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image and reasoning support (including OpenClaw, OpenCode and Cherry)
+ Drag and drop conversation and queued task reordering
+ Pro Yearly plan (2 months free)
2.2.0 Jul 16
+ File Attachments including support for Code, PDF, Word, and ePub
+ Text-only inference of images using metadata, OCR, and feature detection
+ File Tools including Grep, Find and Replace, and Read/Write Operations
+ Caching improvements to reduce misses
+ Distributed compute context precision support
+ Server API reasoning support
+ Improvements to GLM 5.2 and MTP inferencing
+ Support for Ornith-1.0 vision
+ Engine stability fixes
+ Fixed edge case crash and memory leak
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Support for GLM 5.2, GLM 5.2 MTP and MiniMax M3
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Faster persistent prompt caching (up to ~99x for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Server API image support (including OpenClaw and Cherry)
+ Drag and drop conversation reordering
+ Queued task reordering
+ Initial support for macOS 27 beta
+ Pro Yearly plan (2 months free)
2.1.0 Jul 7
+ Support for GLM 5.2, GLM 5.2 MTP and MiniMax M3
+ More engine and MTP stability improvements
+ Server API image support (including OpenClaw and Cherry)
+ Drag and drop conversation reordering
+ Queued task reordering
+ Loop detection and message replay improvements
+ Support for more DeepSeek V4 community models
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Distributed Compute stability improvements
+ Caching improvements for Distributed Compute and Kimi K2.7
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Faster persistent prompt caching (up to ~99x for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Initial support for macOS 27 beta
+ Pro Yearly plan (2 months free)
2.0.6 Jun 29
+ Support for GLM 5.2 and MiniMax M3
+ Engine stability improvements
+ Distributed Compute stability improvements
+ Caching Improvements for Distributed Compute and Kimi K2.7
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Faster persistent prompt caching (up to ~99x for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Initial support for macOS 27 beta
+ Pro Yearly plan (2 months free)
2.0.5 Jun 23
+ Support for GLM 5.2
+ Multimodal inference via server implemented
- Support for arc-1 SAP ADT MCP Tools Server
+ Improved support for messages with images
- Improvements to multiprocessing queues
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Faster persistent prompt caching (up to ~99x for cache hits)
+ MCP Server support available in Tools, including Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Support for M5 Neural Accelerators (~4x prompt processing)
+ Server API response length override, clone multiple chats, faster new conversations, dark mode icon
+ Initial support for macOS 27 beta
+ Pro Yearly plan (2 months free)
2.0.4 Jun 18
+ Support for GLM 5.2
+ Multimodal inference via server implemented
- Support for arc-1 SAP ADT MCP Tools Server
+ Improved support for messages with images
- Improvements to multiprocessing queues
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Faster persistent prompt caching (up to ~99x for cache hits)
+ MCP Server support available in Tools, including Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Support for M5 Neural Accelerators (~4x prompt processing)
+ Server API response length override, clone multiple chats, faster new conversations, dark mode icon
+ Initial support for macOS 27 beta
+ Pro Yearly plan (2 months free)
2.0.3 Jun 17
+ Improvements to new Multiprocessing engine and caching for Server API
+ Validation for Kimi K2.7 Code and GLM 5.1 Code LoRA
+ DeepSeek INF model support for MTP
+ Server API response length override
+ Clone multiple chats
+ Faster new conversations
+ Dark mode icon
+ Initial support for macOS 27 beta
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Multiprocessing engine for inferencing multiple models at the same time
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster, see model description)
+ Faster persistent prompt caching (up to ~99x for cache hits)
+ Improvements to Server API tool calls
+ MCP Server support available in Tools, including Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Support for M5 Neural Accelerators (~4x prompt processing)
+ Pro Yearly plan (2 months free)
2.0.2 Jun 15
+ Multiprocessing engine for inferencing multiple models at the same time
+ Support for Nemotron 3 Ultra, Nemotron Omni, Step 3.7 Flash Vision, Granite Vision, and Qwen3-Omni
+ Fixes for distributed compute detection and DeepSeek V4 verbose thinking
+ MTP Speculative Decoding and Batching for DeepSeek V4
+ Markdown rendering improvements
+ Faster persistent prompt caching
+ Performance improvements for Batching cancellations
+ Improvements to Server API tool calls
+ Multimodal inference end token and cache fixes
+ More bug fixes and performance improvements
Also, in case you missed the last updates:
+ Multi-computer clustering for inferencing larger models
+ Image and Audio inference
+ MTP Speculative Decoding for Qwen3.6 and Gemma4 (up to 2x faster, see model description)
+ Performance improvements for Kimi K2.6, Qwen 3.5 and 3.6 vision
+ MCP Server support available in Tools, including Playwright, MindsDB, Microsoft Learn, and DeepWiki
+ Support for M5 Neural Accelerators (~4x prompt processing)
+ Pro Yearly plan (2 months free)
2.0.1 Jun 7
+ DeepSeek V4.1 cluster support
+ Improvements to multi-node cluster caching
+ GLM 5.3 Flash multimodal cache support
+ Bug fixes and performance improvements
Also, in case you missed the recent updates:
+ Multi-node cluster support for Qwen vision and MiMo V2.6 vision
+ Answer Now button to dynamically end thinking
+ Inference multiple servers at the same time
+ Anthropic API and Claude Code support, including cache optimisations
+ MTP Thinking limit and multiprocessing support
+ Missed tool call detection (enabled by default)
+ HTML/JS validator and Python code viewer
+ File Attachments including support for Code, PDF, Word, and ePub
+ Multi-computer clustering for inferencing larger models
+ MTP and DSpark Speculative Decoding for DeepSeek V4, Gemma4, and Qwen3.6 (up to 2x faster)
+ Persistent prompt caching (enable in Settings - up to ~99x faster for cache hits)
+ MCP Server support including arc-1 SAP, Playwright, MindsDB, Microsoft Learn, and DeepWiki
more Version 2.4.4 20h ago
Data Not Collected The developer does not collect any data from this app.
