v2.0.5

Big-AGI 2.0.5 - Roberto

Named after my father, on his birthday.

Resilience, persistence, excellence.

Long-running AI sessions get a serious upgrade. Anthropic Containers give you persistent files, Skills, and bash sessions that span turns. Gemini Deep Research is now fully resumable - reconnect after pauses, recover from server cuts, cancel cleanly. Add to that Opus 4.7 with 1M tokens, GPT-5.5, DeepSeek V4, and two new vendors.

Models

• Anthropic Opus 4.7 with 1M token context, X-High effort, auto-detect features and 0-day models
• OpenAI GPT-5.5 family with 1M context, plus gpt-image-2 for image generation
• DeepSeek V4 Flash and Pro with native thinking/reasoning_effort
• Gemini 3.1 Flash, 3.1 Flash Lite, 3.1 Flash TTS, Robotics-ER 1.6, and Gemma 4
• xAI Grok 4.3 with encrypted reasoning and reasoning summaries
• Moonshot Kimi K2.6
• New vendors: MiniMax and Arcee AI

Anthropic

• Containers (1h): persistent files, bash sessions, Skills, dynamic filtering, temp files across turns
• Skills: enabled by default with improved selector
• File API: download, embed, delete - text and image inlining via Container -> File pipeline
• Server-side recovered errors ignored, refusal details surfaced, code execution failures traced

Gemini Deep Research

• Full SSE migration (no more POST + GET-loop)
• Resumable across pauses with full recovery mode (#1088)
• Cancel and delete controls, multiple reattaches
• Audio + thoughts support, image-as-figure handling
• Option to disable visualizations (#1095)

Beam

• Auto-title beam chats when done (#1078)
• No accidental wipe with Ctrl+Shift+B (#1079); no re-beam while open
• No ghost columns in 2-beam mode (#1073); title edit fix (#1085)

UX

• Inline message timestamps with TimeAgo (#1065)
• New chat on startup option (#1002)
• Single-dollar inline LaTeX (#795)
• Model pub dates, "new" badges on models under 30 days
• Sort LLM categories by name, filter chat drawer by open beams

Also New

• AIX Resume framework with stop/cancel/recovery across providers
• xAI: encrypted reasoning by default, reasoning summaries always (#1091)
• OpenAI Responses: encrypted reasoning, tool re-parsing fixes
• New Global AI Settings store
• Performance: throttled code highlighting, memoized reasoning chips, lazy per-chat stores

Fixes

• Voice features hidden in browsers without Speech Recognition (#1061)
• OpenRouter models refresh (#804)
• Zod 4 type strictness (#1072)

New Contributors

• @alexdenton123 made their first contribution in #1062
• @tredondo made their first contribution in #1074

Full Changelog: v2.0.4...v2.0.5

Get Started

Available now at big-agi.com, via Docker, or self-host from source.

v2.0.4

Big-AGI 2.0.4 - Hyper Params

Precision matters.

This release is about getting models right: accurate parameter ranges for reasoning effort, verbosity, and temperature across providers and through 3rd parties such as OpenRouter and AWS. Plus the latest frontier models, a new cloud platform, and Anthropic going deeper.

A huge shout-out goes to @Slava-256, for making Big-AGI's model support the best-in-class with this release. From filing detailed issues for accurate reasoning effort ranges (#940, #944), verbosity accuracy (#947, #948), temperature support (#953), to pushing for Opus 4.6 support (#962), and many more (31 closed tickets!). Slava's thorough testing, precise reports, findings, and reference code are the reason why 2.0.4 was focused on precise parameters support.

Models & Parameters

• Anthropic Opus 4.6 with adaptive thinking and 1M tokens, Sonnet 4.6 and new features
• Google Gemini 3.1 Pro, Nano Banana 2
• OpenAI GPT-5.4 family with 1M context and reasoning controls
• xAI Grok 4.20
• New Z.ai GLM models support, with configurable thinking
• API-validated parameters: reasoning effort, budget, verbosity, temperature, tools, are tested on live APIs
• Model-specific controls: each model exposes only the parameter values it actually supports
• Automated sweep testing across native APIs keeps parameter definitions current

Anthropic API mastery

• Configurable web search and web fetch depth, and dynamic search
• API controls such as US-based inference and Fast mode for Opus (2.5x speed, at 6x cost)
• Continuation for long outputs (pause_turn support)

AWS Bedrock support (New)

• Authentication via both IAM credentials, or using a single AWS Bedrock token
• Native support for 3 native AWS Bedrock protocols: Anthropic models via Invoke, Amazon Nova models via Converse, and other models via OpenAI-compatible ChatCompletions

Also New

• Attachments on any message, lossless images option (warning: can be very large on Gemini models)
• Rich text copy, reasoning trace controls
• System theme mode follow
• Mobile focus mode

New Contributors

• @Blacksuan19 made their first contribution (SSR guard, system theme mode) in #1011
• @dLo999 made their first contribution (local time for backup exports) in #1041
• Full Changelog: v2.0.3...v2.0.4

Get Started

Available now at big-agi.com, via Docker, or self-host from source.

v2.0.3

Big-AGI 2.0.3 - Red Carpet

What's New

Models updates

This release embraces models precision: improved parameter configuration to match the model vendors, a new native xAI-flavored responses API, a new model vendor, and all the newest models including the already legendary Kimi 2.5, great for code.

• Kimi K2.5 with image understanding and reasoning
• Gemini 3 Flash with 4-level thinking (high, medium, low, minimal)
• GPT 5.2 with x-high reasoning and native search
• GPT Image 1.5 for image generation
• Novita.ai full provider support with automatic capability detection
• LMStudio model capability support and images
• xAI Responses native protocol support with X post search and code execution

Also New

• Google Drive integration - attach docs, sheets, images with optimal LLM conversion
• Inworld speech synthesis with streaming playback
• Model duplication for custom configurations
• AI Injector, collapsible model groups, PDF auto-OCR
• Copy as-seen, reorder messages in cleanup mode
• Anthropic and OpenAI API fixes, security patches, CJK/IME input

Full Changelog: v2.0.2...v2.0.3

Get Started

Available now at big-agi.com, via Docker, or self-host from source.

v2.0.2

Big-AGI 2.0.2 - Speex

What's New

Speex: Multi-Vendor Speech Synthesis

A complete redesign of text-to-speech with support for multiple providers:

• Web Speech API - Zero-config browser-native voices (works offline)
• LocalAI - Self-hosted TTS with Kokoro, Bark, Piper backends
• ElevenLabs - High-quality voice synthesis and cloning
• OpenAI TTS - gpt-4o-mini-tts with voice instructions, plus tts-1 and tts-1-hd

Engines auto-detect from your configured LLM services—no duplicate API key entry. Priority-based fallback ensures speech always works.

Also New

• Claude Opus 4.5 full support
• Starred models for quick access
• Continued stability improvements

Full Changelog: v2.0.1...v2.0.2

Get Started

Available now at big-agi.com, via Docker, or self-host from source.

Open v2.0.1

Big-AGI 2.0.1 - Heavy Critters

What's New

Latest Models

• Gemini 3 Pro with code execution support
• Nano Banana Pro for controllable image generation
• Grok 4.1 Fast, the GPT-5.1 family, Kimi K2 Thinking
• Updated model definitions across all 17+ providers

Enhanced Features

• Image Generation now supports Azure and LocalAI alongside OpenAI
• OpenRouter auto-detects model capabilities and reasoning
• Call transcripts saved automatically
• Generate persona images
• Search functionality in Beam

Improvements

• Enhanced error resilience (85% reduction in connection issues)
• 280 commits of fixes and optimizations
• AI-Automatic feature development for community contributions

Full Changelog: v2.0.0...v2.0.1

Get Started

Available now at big-agi.com, via Docker, or self-host from source.

v2.0.0

🚀 Big-AGI 2.0.0 - Open, Free and Pro

A year in the making, redesigned for professionals.

Big-AGI 2.0 is here. More than an update; we've rebuilt core components, massively improved the UI, and are officially launching the three-tier model to support the project's next phase.

Built for Engineers architecting systems, Founders making decisions, Researchers validating hypotheses, and IT departments deploying the top value to their company, and after thousands of hours of engineering, we're delivering what you asked for: a professional-grade AI workspace that's local-first, vendor-neutral, and madly optimized for zero latency.

What's New

Beam 2: Multi-Modal Intelligence

Program-based reasoning across multiple models simultaneously, with images, follow-ups, and saved presets.

Professional Features

• Reasoning Models: Anthropic Claude Opus 4.1, Gemini 2.5 Pro, OpenAI GPT-5/o3/o1, xAI Grok 4, etc.
• Web Search native and with citations for supported models
• Image Generation with Nano Banana and gpt-image-1, including edits
• Local FileSystem integration: drag/drop files to add context, diff them, and save them back, automatically
• UI completely overhauled for desktop and mobile

Cloud Sync (Pro Only)

10 months of solo engineering. Real-time sync across unlimited devices, 1GB storage. Your data, your control.

Technical Improvements

• Local-first redesign: Everything runs in your browser when possible. Zero latency. No vendor lock-in.
• Multi-part/multi-modal almost every subsystem has been redesigned and drastically improved
• And plenty more - see the latest on big-agi.com/changes

Three Tiers

• Big-AGI Open (self-host): First to get new models. Maximum control. IT departments love it.
• Big-AGI Free (big-agi.com): Full core experience with Beam, Personas, and the best UX
• Big-AGI Pro (from $10.99/mo): Everything + Cloud Sync across unlimited devices with 1GB storage

As an independent, non-VC-funded project, Pro subscriptions fund all development including the free and open-source tiers.

Inception Screenshot

New Contributors

Welcome and thank you to our 12 new contributors who made their first contributions to Big-AGI! Your code is now part of something professionals worldwide rely on daily.

• @powyncify made their first contribution (UI improvements) in #830
• @sam0jones0 made their first contribution (Claude thinking models) in #811
• @darthalex2014 made their first contribution (multi-key support) in #777
• @Jonathan523 made their first contribution (Gemini custom endpoints) in #716
• @jayrepo made their first contribution (Docker SSL fixes) in #706
• @matrix303 made their first contribution (Docker Node 22) in #695
• @darshmashru made their first contribution (documentation fix) in #629
• @mapringg made their first contribution (DeepSeek vendor + Mac shortcuts) in #580
• @therippa made their first contribution (Personas panel fix) in #575
• @JeremiLorenti made their first contribution (DALL-E settings) in #572
• @blakkd made their first contribution (LocalAI URL fix) in #570
• @pichuang made their first contribution (Kubernetes support) in #152

Full Changelog: v1.16.8...v2.0.0

For Institutions

Need shared prompts, SSO, or managed deployments? Contact enrico@big-agi.com - we're actively building enterprise features based on research group requirements.

In Summary

Big-AGI 2.0 represents our commitment to building the AI workspace that experts deserve. No compromises on speed, no vendor lock-in, no BS. Just a powerful tool that works exactly how you need it to.

As an independent, non-VC-funded project, Pro subscriptions directly fund all development. Early subscribers shape the roadmap.

Upgrade today via big-agi.com, Docker, or self-host from source.

v1.16.0

🚀 big-AGI 1.16.0 - Crystal Clear

Clarity and transparency in your AI interactions.

big-AGI 1.16.0 - Crystal Clear - brings enhancements and new features designed to ensure that every interaction with big-AGI is smoother, more intuitive, and simply better - as we always do. We hope you'll love the updated models across the board, and the new cost tracking (please refresh the models to load the latest prices, now supported on Anthropic, OpenAI, and OpenRouter), which will translate to saving money, as anything that can be measured can be optimized for.

What's Changed

• Chat Cost Estimation: display a cost estimation for supported networks*. hover price for token info
  
• Enhanced Beam based on your feedback: auto-selection and remembering networks and UX simplifications #470
• Major improvements to the Auto-Diagrams Tool
• Save/Load Chat Files with Ctrl+S and Ctrl+O #466
• @joriskalz: YouTube Transcriber persona: transcribe and chat with YouTube videos (un-hide it from the selector) #500
• @mludvig: 3x faster on Apple silicon with Docker multi-arch builds #507
• More: Perfect formulas/LaTeX rendering #508, soft-wrap code #517, chat text selection toolbar, dark mode fixes #520
• Doc corrections by @dandv and @PrivTEC and in #521 #522 #525
• Updated Anthropic*, Groq, Ollama, OpenAI*, OpenRouter* and Perplexity models
• For Developers: updated LLM data structures
• Full Changelog

New Contributors

• A warm welcome and thank you to @mludvig, @PrivTEC, and @dandv for their first contributions to big-AGI!

In Summary

Transparency and user experience improvements, for an improved quality of life. We are thankful to be able to merge contributions for the community, including a new YouTube Transcriber persona and the faster multi-arch (ARM, x86) builds.

Upgrade today and experience the clearer, faster, and more intuitive big-AGI. Available via code, Docker, or on the free instance on big-agi.com. 🚀

v1.15.1

Minor release

• Support for the Google 1.5 Pro models, just released
• Support for the OpenAI 2024-04-09 models, just released
• Improved saving/loading of Desktop chats (Ctrl + S, Ctrl + O)
• Misc: added docs/installation.md, improver

New Contributors

• @lunamidori5 made their first contribution in #484
• @dogmatic69 made their first contribution in #497

Full Changelog: v1.15.0...v1.15.1

v1.15.0

🚀 big-AGI 1.15.0 - Beam

Introducing multi-model AI chats.

We're thrilled to announce big-AGI 1.15.0 - Beam. Beam is a new chat modality in big-AGI that empowers users to find better answers by leveraging the collective intelligence of multiple models. It's a game-changer for brainstorming, decision-making, and creativity.

The word "groundbreaking" gets tossed around a lot these days, but this is new technique really is, and this is a world exclusive for you big-AGI lovers. With over 400 commits and 10,000+ lines of code changed, big-AGI 1.15 sets a milestone in AI chats.

• Beam: Unveiling the multi-model AI chat experience. #470
• Managed Deployments Auto-Configuration: Simplify your setup with auto-configuration for managed deployments: set env vars on the backend and the frontend will pick up all the models automatically - works well local models too, such as LocalAI. #436
• Message Starring ⭐ and Filtering: Star important messages, and in the future you'll be able to attach and filter chats. It's already working, but we're still testing the UI patterns for it #476
• Persona Enhancements: The default AI persona just got better
• UI and Model Fixes: Ironed out issues with Gemini models and SVGs and polished the UI and icons for a smoother experience

For Developers:

• Imperative LLM Models Discovery: moved from React to imperative code, with reactive hooks. Enables #436
• Full changelog: v1.14.1...v1.15.0

In Summary

1.15.0 - Beam is the most impactful big-AGI release to date. By leveraging the power of multiple models, Beam enables users to explore diverse perspectives, synthesize optimal responses, and push the boundaries of what's possible with AI-assisted brainstorming, decision-making, and creativity.

Experience the future of AI chat and upgrade to Beam today, available via code, Docker, or directly on big-agi.com.

Join our vibrant AGI community on Discord and be part of the conversation! Your feedback and contributions are invaluable in shaping the future of big-AGI. Now, let's embark on this exciting journey together with Beam 🚀

v1.14.0

🚀 big-AGI 1.14.0 - Modelmorphic

The new shape of intelligence and performance.

We're releasing big-AGI 1.14.0, "Modelmorphic." This update expands models support, focuses on improving performance, and delivers user experience enhancements. With over 230 commits and over 8,000 lines of code changed, it's designed to make big-AGI smarter, faster, more reliable.

• (1.14.1) Added support for *Anthropic Claude-3 models
• Added support for Perplexity, including Sonar Online models, and Groq, including models with up to 500 tok/s speeds. Contribution by @Penagwin.
• Added Mistral Large models, and Google Gemini 1.5 models.
• Deeper LocalAI integration with support for model galleries and keys.
• Performance Optimizations: comprehensive improvements for faster operation, lower power consumption, and better memory use
• UI and Usability: updated navigation, search, scaling, shortcuts, and chat interactions for a smoother experience.
• Addressed issues with image warnings, standalone LaTeX blocks, hidden personas in the selector, and more.

For Developers:

• Customizable PlantUML Server via an environment variable, thanks to @fredliubojin.
• New customization documentation and analytics support are now included.
• Added /info/debug for better diagnostics.
• Introduced a "controller-based" chat architecture to simplify per-chat data management, including ephemerals.
• Full changelog: v1.13.1...v1.14.0

New Contributors

Special thanks to our new contributors @Penagwin and @jacksongoode. @Penagwin work has been instrumental for making this release a success.

In Summary

Version 1.14.0 - Modelmorphic represents a leap forward in our commitment to providing advanced, reliable, and user-centric AI tools. We continue to eschew fluff in favor of substance, focusing on what truly matters to our users and developers.

Explore the new features and improvements and upgrade to Modelmorphic today, available via code, Docker, or directly on big-agi.com.

Join our AGI community on Discord and get involved with us! Your feedback and contributions are what drive us forward. Now enjoy Modelmorphic 🚀