aider
92 versions indexed · latest v0.86.0 on Aug 9, 2025
202516
Added support for all GPT-5 models. · Added support for Grok-4 via xai/grok-4 and openrouter/x-ai/grok-4 model names. · Added support for gemini/gemini-2.5-flash-lite-preview-06-17 model, by Tamir Zahavi-Brunner.
Added support for new Gemini models including gemini-2.5-pro, gemini-2.5-flash, and gemini-2.5-pro-preview-06-05 with thinking tokens support.
Added support for new Claude models including the Sonnet 4 and Opus 4 series (e.g., claude-sonnet-4-20250514
Added support for gemini-2.5-pro-preview-05-06 models. · Added support for qwen3-235b models. · Added repo-map support for OCaml and OCaml interface files, by Andrey Popp.
Support for GPT 4.1, mini and nano. · Improved support for using architect mode with Gemini 2.5 Pro. · Add support for xai/grok-3-beta, xai/grok-3-mini-beta, openrouter/x-ai/grok-3-beta
Added support for the openrouter/openrouter/quasar-alpha model. · Currently free on OpenRouter. · Quasar scored 55% on the polyglot coding benchmark · Run with aider --model quasar
OpenRouter OAuth integration · Offer to OAuth against OpenRouter if no model and keys are provided. · Select OpenRouter default model based on free/paid tier status if OPENROUTER_API_KEY is set and no model is specified.
Added support for SOTA Gemini 2.5 Pro. · Added support for DeepSeek V3 0324. · Added a new /context command that automatically identifies which files need to be edited for a given request.
Added support for thinking tokens for OpenRouter Sonnet 3.7. · Added commands to switch between model types: /editor-model for Editor Model, and /weak-model for Weak Model, by csala.
Big upgrade in programming languages supported by adopting tree-sitter-language-pack. · 130 new languages with linter support. · 20 new languages with repo-map support.
Improved support for thinking/reasoning models · Added --thinking-tokens CLI option to control token budget for models that support thinking. · Display thinking/reasoning content from LLMs which return it.
Basic support for Claude 3.7 Sonnet · Use --model sonnet to use the new 3.7 · Thinking support coming soon. · Bugfix to /editor command.
Dynamically changes the Ollama context window to hold the current chat. · Better support for o3-mini, DeepSeek V3 & R1, o1-mini, o1 via secondary API providers.
Full support for o3-mini: aider --model o3-mini · New --reasoning-effort argument: low, medium, high. · Improved handling of context window size limits, with better messaging and Ollama-specific guidance.
Support for DeepSeek R1. · Use shortcut: --model r1 · Also via OpenRouter: --model openrouter/deepseek/deepseek-r1 · Added Kotlin syntax support to repo map, by Paul Walker.
Prompts to help DeepSeek work better when alternating between /ask and /code. · Streaming pretty LLM responses is smoother and faster for long replies. · Streaming automatically turns of for model that don't support it
202451
Full support for o1 models. · Watch files now honors --subtree-only, and only watches that subtree. · Improved prompting for watch files, to work more reliably with more models.
Watch files improvements · Use # ... AI? comments to trigger aider and ask questions about your code. · Now watches all files, not just certain source files.
Aider works with LLM web chat UIs. · New --copy-paste mode. · New /copy-context command. · Set API keys and other environment variables for all providers from command line or yaml conf file.
Use aider in your IDE or editor. · Run aider --watch-files and it will watch for instructions you add to your source files. · One-liner # ... or // ... comments that start or end with 'AI' are instructions to aider.
PDF support for Sonnet and Gemini models. · Added --voice-input-device to select audio input device for voice recording · Added --timeout option to configure API call timeouts.
Added --alias config to define custom model aliases. · Added --[no-]detect-urls flag to disable detecting and offering to scrape URLs found in the chat. · Ollama models now default to an 8k context window.
Added /editor command to open system editor for writing prompts · Full support for gpt-4o-2024-11-20. · Stream o1 models by default.
Support for Qwen 2.5 Coder 32B. · /web command just adds the page to the chat, without triggering an LLM response. · Improved prompting for the user's preferred chat language. · Improved handling of LiteLLM exceptions.
Full support for Claude 3.5 Haiku · Scored 75% on aider's code editing leaderboard. · Almost as good as Sonnet at much lower cost. · Launch with --haiku to use it.
Load and save aider slash-commands to files · /save command will make a file of /add and /read-only commands that recreate the current file context in the chat. · /load will replay the commands in the file.
Full support for Sonnet 10/22, the new SOTA model on aider's code editing benchmark. · Aider uses Sonnet 10/22 by default. · Improved formatting of added and read-only files above chat prompt
Improvements to /read-only · Now supports shell-style auto-complete of the full file system. · Still auto-completes the full paths of the repo files like /add. · Now supports globs like src//.py
Use a pair of Architect/Editor models for improved coding · Use a strong reasoning model like o1-preview as your Architect. · Use a cheaper, faster model like gpt-4o as your Editor.
Support for OpenAI o1 models · o1-preview now works well with diff edit format. · o1-preview with diff now matches SOTA leaderboard result with whole edit format. · aider --model o1-mini
Enables prompt caching for Sonnet via OpenRouter · Enables 8k output tokens for Sonnet via VertexAI and DeepSeek V2.5. · New /report command to open your browser with a pre-populated GitHub Issue.
Offer to submit a GitHub issue pre-filled with uncaught exception info. · Numerous corner case bug fixes submitted via new pre-filled crash report feature. · New /settings command to show active settings.
Added model settings for gemini/gemini-1.5-pro-exp-0827 and
Keep your prompt cache from expiring with --cache-keepalive-pings. · Pings the API every 5min to keep the cache warm. · You can now bulk accept/reject a series of add url and run shell confirmations.
Aider now offers to run shell commands · Launch a browser to view updated html/css/js. · Install new dependencies. · Run DB migrations.
Prompt caching for Anthropic models with --cache-prompts. · Caches the system prompt, repo map and /read-only files. · Repo map recomputes less often in large/mono repos or when caching enabled.
Infinite output for DeepSeek Coder, Mistral models in addition to Anthropic's models. · New --deepseek switch to use DeepSeek Coder. · DeepSeek Coder uses 8k token output.
Add read-only files to the chat context with /read and --read, including from outside the git repo. · /diff now shows diffs of all changes resulting from your request, including lint and test fixes.
Performance improvements for large/mono repos. · Added --subtree-only to limit aider to current directory subtree. · Should help with large/mono repo performance.
Commit message improvements · Added Conventional Commits guidelines to commit message prompt. · Added --commit-prompt to customize the commit message prompt.
New /ask command to ask about your code, without making any edits. · New /chat-mode command to switch chat modes · ask: Ask questions about your code without making any changes.
GPT 4o mini scores like the original GPT 3.5 on aider's code editing benchmark, using whole edit format. · Aider is better at offering to add files to the chat on Windows.
Default pip install size reduced by 3-12x. · Added 3 package extras, which aider will offer to install when needed · aider-chat[help] · aider-chat[browser]
Use /help to ask for help about using aider, customizing settings
Performance release · 5X faster launch! · Faster auto-complete in large git repos (users report 100X speedup)!
Allow Claude 3.5 Sonnet to stream back >4k tokens! · It is the first model capable of writing such large coherent, useful code edits. · Do large refactors or generate multiple files of new code in one go.
Improved prompting to discourage Sonnet from wasting tokens emitting unchanging code (#705). · Improved error info for token limit errors. · Options to suppress adding '(aider)' to the git author and committer names.
Use --sonnet for Claude 3.5 Sonnet, which is the top model on aider's LLM code editing leaderboard. · All AIDER_xxx environment variables can now be set in .env (by @jpshack-at-palomar).
Use --vim for vim keybindings in the chat. · Add LLM metadata via .aider.models.json file (by @caseymcc). · More detailed error messages on token limit errors.
Repo map is now optimized based on text of chat history as well as files added to chat. · Improved prompts when no files have been added to chat to solicit LLM file suggestions.
Aider can now lint your code and fix any errors. · Aider automatically lints and fixes after every LLM edit. · You can manually lint-and-fix files with /lint in the chat or --lint on the command line.
Aider now uses GPT-4o by default. · GPT-4o tops the aider LLM code editing leaderboard at 72.9%, versus 68.4% for Opus. · GPT-4o takes second on aider's refactoring leaderboard with 62.9%, versus Opus at 72.3%.
Updated prompting to use more natural phrasing about files, the git repo, etc. Removed reliance on read-write/read-only terminology. · Refactored prompting to unify some phrasing across edit formats.
Added native support for Deepseek models using DEEPSEEK_API_KEY and deepseek/deepseek-chat, etc rather than as a generic OpenAI compatible API.
Aider LLM code editing leaderboards that rank popular models according to their ability to edit code. · Leaderboards include GPT-3.5/4 Turbo, Opus, Sonnet, Gemini 1.5 Pro, Llama 3, Deepseek Coder & Command-R+.
Aider is now also AI pair programming in your browser! Use the --browser switch to launch an experimental browser based version of aider.
Added Gemini 1.5 Pro as a recommended free model. · Added better support for Groq's free Llama3 70B model. · Allow repo map for 'whole' edit format. · Added --models to search the available models.
Added support for directly connecting to Anthropic, Cohere, Gemini and many other LLM providers. · Added --weak-model which allows you to specify which model to use for commit messages and chat history summarization.
Added support for new gpt-4-turbo-2024-04-09 and gpt-4-turbo models.
Improved repomap support for typescript · Bugfix: Only /undo the files which were part of the last commit, don't stomp other dirty files · Bugfix: Show clear error message when OpenAI API key is not set.
Use GPT-4 Turbo by default. · Added -3 and -4 switches to use GPT 3.5 or GPT-4 (non-Turbo). · Bug fix to avoid reflecting local git errors back to GPT. · Improved logic for opening git repo on launch.
New /web command which scrapes the url, turns it into fairly clean markdown and adds it to the chat. · Updated all OpenAI model names, pricing info · Default GPT 3.5 model is now gpt-3.5-turbo-0125.
Added support for --model gpt-4-0125-preview and OpenAI's alias --model gpt-4-turbo-preview. The --4turbo switch remains an alias for --model gpt-4-1106-preview at this time.
Improvements for unified diff editing format. · Added ! as an alias for /run. · Autocomplete for /add and /drop now properly quotes filenames with spaces. · The /undo command asks GPT not to just retry reverted edit.
Support for python 3.12. · Improvements to unified diff editing format. · New --check-update arg to check if updates are available and exit with status code.
Add images to the chat to automatically use GPT-4 Vision · Bugfixes · Improved unicode encoding for /run command output · Prevent false auto-commits on Windows
202325
Significantly reduced 'lazy' coding from GPT-4 Turbo due to new unified diff edit format · Score improves from 20% to 61% on new 'laziness benchmark'. · Aider now uses unified diffs by default for gpt-4-1106-preview.
Improved prompting for both GPT-4 and GPT-4 Turbo. · Far fewer edit errors from GPT-4 Turbo (gpt-4-1106-preview).
Support for OpenAI's new 11/06 models · gpt-4-1106-preview with 128k context window · gpt-3.5-turbo-1106 with 16k context window · Benchmarks for OpenAI's new 11/06 models
Updated tree-sitter dependencies to streamline the pip install process
Improved repository map using tree-sitter · Switched from 'edit block' to 'search/replace block', which reduced malformed edit blocks. Benchmarked at 66.2%, no regression.
Added support for .aiderignore file, which instructs aider to ignore parts of the git repo. · New --commit cmd line arg, which just commits all pending changes with a sensible commit message geneated by gpt-3.5.
Easily run aider from a docker image · Fixed bug with chat history summarization. · Fixed bug if soundfile package not available.
/add and /drop handle absolute filenames and quoted filenames · /add checks to be sure files are within the git repo (or root) · If needed, warn users that in-chat file paths are all relative to the git repo
Support for Claude2 and other LLMs via OpenRouter · Documentation for running the aider benchmarking suite · Aider now requires Python >= 3.9
Only git commit dirty files that GPT tries to edit · Send chat history as prompt/context for Whisper voice transcription · Added --voice-language switch to constrain /voice to transcribe to a specific language
Voice-to-code support, which allows you to code with your voice. · Fixed bug where /diff was causing crash. · Improved prompting for gpt-4, refactor of editblock coder.
Added a progress bar when initially creating a repo map. · Fixed bad commit message when adding new file to empty repo. · Fixed corner case of pending chat history summarization when dirty committing.
Automatically summarize chat history to avoid exhausting context window. · More detail on dollar costs when running with --no-stream
/add and /drop always use paths relative to the git root · Encourage GPT to use language like 'add files to the chat' to ask users for permission to edit them.
Added /git command to run git from inside aider chats. · Use Meta-ENTER (Esc+ENTER in some environments) to enter multiline chat messages.
Support for the OpenAI models in Azure · Added --show-repo-map · Improved output when retrying connections to the OpenAI API · Redacted api key from --verbose output
Added --dark-mode and --light-mode to select colors optimized for terminal background · Install docs link to NeoVim plugin · Reorganized the --help output
Disabled general availability of gpt-4 (it is rolling out, not 100% available yet)
Ask to create a git repo if none found, to better track GPT's code changes · Glob wildcards are now supported in /add and /drop commands · Pass --encoding into ctags, require it to return utf-8
Benchmark comparing code editing in GPT-3.5 and GPT-4 · Improved Windows support · Fixed bugs related to path separators in Windows · Added a CI step to run all tests on Windows
Fixed a bug to allow aider to edit files that contain triple backtick fences.
Fixed a bug in the display of streaming diffs in GPT-3.5 chats
Graceful handling of context window exhaustion, including helpful tips. · Added --message to give GPT that one instruction and then exit after it replies and any edits are performed.
Support for gpt-3.5-turbo-16k, and all OpenAI chat models · Improved ability to correct when gpt-4 omits leading whitespace in code edits · Added --openai-api-base to support API proxies, etc.
Added support for gpt-3.5-turbo and gpt-4-32k. · Added --map-tokens to set a token budget for the repo map, along with a PageRank based algorithm for prioritizing which files and identifiers to include in the map.