- Models hosted on [Replicate](https://replicate.com/) can now be accessed using the [llm-replicate](https://github.com/simonw/llm-replicate) plugin, including the new Llama 2 model from Meta AI. More details here: [Accessing Llama 2 from the command-line with the llm-replicate plugin](https://simonwillison.net/2023/Jul/18/accessing-llama-2/).
- Model providers that expose an API that is compatible with the OpenAPI API format, including self-hosted model servers such as [LocalAI](https://github.com/go-skynet/LocalAI), can now be accessed using {ref}`additional configuration <openai-compatible-models>` for the default OpenAI plugin. [#106](https://github.com/simonw/llm/issues/106)
- OpenAI models that are not yet supported by LLM can also {ref}`be configured <openai-extra-models>` using the new `extra-openai-models.yaml` configuration file. [#107](https://github.com/simonw/llm/issues/107)
- The {ref}`llm logs command <viewing-logs>` now accepts a `-m model_id` option to filter logs to a specific model. Aliases can be used here in addition to model IDs. [#108](https://github.com/simonw/llm/issues/108)
- Logs now have a SQLite full-text search index against their prompts and responses, and the `llm logs -q SEARCH` option can be used to return logs that match a search term. [#109](https://github.com/simonw/llm/issues/109)
- [llm-gpt4all](https://github.com/simonw/llm-gpt4all) adds support for 17 models that can download and run on your own device, including Vicuna, Falcon and wizardLM.
- [llm-mpt30b](https://github.com/simonw/llm-mpt30b) adds support for the MPT-30B model, a 19GB download.
- [llm-palm](https://github.com/simonw/llm-palm) adds support for Google's PaLM 2 via the Google API.
A comprehensive tutorial, {ref}`writing a plugin to support a new model <tutorial-model-plugin>` describes how to add new models by building plugins in detail.
- {ref}`python-api` documentation for using LLM models, including models from plugins, directly from Python. [#75](https://github.com/simonw/llm/issues/75)
- Messages are now logged to the database by default - no need to run the `llm init-db` command any more, which has been removed. Instead, you can toggle this behavior off using `llm logs off` or turn it on again using `llm logs on`. The `llm logs status` command shows the current status of the log database. If logging is turned off, passing `--log` to the `llm prompt` command will cause that prompt to be logged anyway. [#98](https://github.com/simonw/llm/issues/98)
- New database schema for logged messages, with `conversations` and `responses` tables. If you have previously used the old `logs` table it will continue to exist but will no longer be written to. [#91](https://github.com/simonw/llm/issues/91)
- New `-o/--option name value` syntax for setting options for models, such as temperature. Available options differ for different models. [#63](https://github.com/simonw/llm/issues/63)
-`llm models list --options` command for viewing all available model options. [#82](https://github.com/simonw/llm/issues/82)
-`llm "prompt" --save template` option for saving a prompt directly to a template. [#55](https://github.com/simonw/llm/issues/55)
- Prompt templates can now specify {ref}`default values <prompt-default-parameters>` for parameters. Thanks, Chris Mungall. [#57](https://github.com/simonw/llm/pull/57)
-`llm openai models` command to list all available OpenAI models from their API. [#70](https://github.com/simonw/llm/issues/70)
-`llm models default MODEL_ID` to set a different model as the default to be used when `llm` is run without the `-m/--model` option. [#31](https://github.com/simonw/llm/issues/31)
-`llm install -e directory` command for installing a plugin from a local directory.
- The `LLM_USER_PATH` environment variable now controls the location of the directory in which LLM stores its data. This replaces the old `LLM_KEYS_PATH` and `LLM_LOG_PATH` and `LLM_TEMPLATES_PATH` variables. [#76](https://github.com/simonw/llm/issues/76)
{ref}`prompt-templates` is a new feature that allows prompts to be saved as templates and re-used with different variables.
Templates can be created using the `llm templates edit` command:
```bash
llm templates edit summarize
```
Templates are YAML - the following template defines summarization using a system prompt:
```yaml
system: Summarize this text
```
The template can then be executed like this:
```bash
cat myfile.txt | llm -t summarize
```
Templates can include both system prompts, regular prompts and indicate the model they should use. They can reference variables such as `$input` for content piped to the tool, or other variables that are passed using the new `-p/--param` option.
This example adds a `voice` parameter:
```yaml
system: Summarize this text in the voice of $voice
```
Then to run it (via [strip-tags](https://github.com/simonw/strip-tags) to remove HTML tags from the input):
> My previous test subject seemed to have learned something new about iMovie. They exported keynote slides as individual images [...] Quite impressive for a human.
The {ref}`prompt-templates` documentation provides more detailed examples.
### Continue previous chat
You can now use `llm` to continue a previous conversation with the OpenAI chat models (`gpt-3.5-turbo` and `gpt-4`). This will include your previous prompts and responses in the prompt sent to the API, allowing the model to continue within the same context.
Use the new `-c/--continue` option to continue from the previous message thread:
```bash
llm "Pretend to be a witty gerbil, say hi briefly"
```
> Greetings, dear human! I am a clever gerbil, ready to entertain you with my quick wit and endless energy.
```bash
llm "What do you think of snacks?" -c
```
> Oh, how I adore snacks, dear human! Crunchy carrot sticks, sweet apple slices, and chewy yogurt drops are some of my favorite treats. I could nibble on them all day long!
The `-c` option will continue from the most recent logged message.
To continue a different chat, pass an integer ID to the `--chat` option. This should be the ID of a previously logged message. You can find these IDs using the `llm logs` command.
Thanks [Amjith Ramanujam](https://github.com/amjith) for contributing to this feature. [#6](https://github.com/simonw/llm/issues/6)
### New mechanism for storing API keys
API keys for language models such as those by OpenAI can now be saved using the new `llm keys` family of commands.
To set the default key to be used for the OpenAI APIs, run this:
```bash
llm keys set openai
```
Then paste in your API key.
Keys can also be passed using the new `--key` command line option - this can be a full key or the alias of a key that has been previously stored.
See link-to-docs for more. [#13](https://github.com/simonw/llm/issues/13)
### New location for the logs.db database
The `logs.db` database that stores a history of executed prompts no longer lives at `~/.llm/log.db` - it can now be found in a location that better fits the host operating system, which can be seen using:
```bash
llm logs path
```
On macOS this is `~/Library/Application Support/io.datasette.llm/logs.db`.
To open that database using Datasette, run this:
```bash
datasette "$(llm logs path)"
```
You can upgrade your existing installation by copying your database to the new location like this:
```bash
cp ~/.llm/log.db "$(llm logs path)"
rm -rf ~/.llm # To tidy up the now obsolete directory
```
The database schema has changed, and will be updated automatically the first time you run the command.
That schema is [included in the documentation](https://llm.datasette.io/en/stable/logging.html#sql-schema). [#35](https://github.com/simonw/llm/issues/35)
### Other changes
- New `llm logs --truncate` option (shortcut `-t`) which truncates the displayed prompts to make the log output easier to read. [#16](https://github.com/simonw/llm/issues/16)
- Documentation now spans multiple pages and lives at <https://llm.datasette.io/> [#21](https://github.com/simonw/llm/issues/21)
- Default `llm chatgpt` command has been renamed to `llm prompt`. [#17](https://github.com/simonw/llm/issues/17)
- Removed `--code` option in favour of new prompt templates mechanism. [#24](https://github.com/simonw/llm/issues/24)
- Responses are now streamed by default, if the model supports streaming. The `-s/--stream` option has been removed. A new `--no-stream` option can be used to opt-out of streaming. [#25](https://github.com/simonw/llm/issues/25)
- The `-4/--gpt4` option has been removed in favour of `-m 4` or `-m gpt4`, using a new mechanism that allows models to have additional short names.
- The new `gpt-3.5-turbo-16k` model with a 16,000 token context length can now also be accessed using `-m chatgpt-16k` or `-m 3.5-16k`. Thanks, Benjamin Kirkbride. [#37](https://github.com/simonw/llm/issues/37)
- Improved display of error messages from OpenAI. [#15](https://github.com/simonw/llm/issues/15)
-`llm logs` command for browsing logs of previously executed completions. [#3](https://github.com/simonw/llm/issues/3)
-`llm "Python code to output factorial 10" --code` option which sets a system prompt designed to encourage code to be output without any additional explanatory text. [#5](https://github.com/simonw/llm/issues/5)
- Tool can now accept a prompt piped directly to standard input. [#11](https://github.com/simonw/llm/issues/11)
- If a SQLite database exists in `~/.llm/log.db` all prompts and responses are logged to that file. The `llm init-db` command can be used to create this file. [#2](https://github.com/simonw/llm/issues/2)
