Github Models support, PDF support for Sonnet, r1, and more

What's Changed

• Add pdf input for Sonnet by @ultronozm in #142
• Add GitHub Models support by @gs-101 in #113
• Add deepseek r1 to the model list by @ahyatt in #146
• Simplify github models to be based of llm-azure by @ahyatt in #147
• Use "chat-model" as the name for "llm-open-compatible" by @whhone in #148
• Accept lists as valid non-standard-params, but prefer vectors by @ahyatt in #149

New Contributors

• @gs-101 made their first contribution in #113
• @whhone made their first contribution in #148

Full Changelog: 0.22.0...0.23.0

Improvements to llm-make-tool, rename from llm-make-tool-function

What's Changed

• Rename llm-tool-functions to llm-tool, fixes to elisp-to-tool by @ahyatt in #141

Tool use interface change, more Claude functionality

What's Changed

• Corrected typos in Snowflake model names and symbols by @s-kostyaev in #135
• Introduce llm-models-add, plus fix an issue with gemini 2.0 model by @ahyatt in #136
• Change function calling methods to tool-use, modify input format, add Claude image and streaming tool use by @ahyatt in #133

Full Changelog: 0.20.0...0.21.0

JSON-mode with schema following, new models

What's Changed

• Fix url by @jsntn in #118
• Add Gemini 2.0 flash to the list of Gemini models by @ahyatt in #121
• Add llama-3.3 and QwQ models by @ahyatt in #122
• Add ability to get JSON object as a specific schema by @ahyatt in #123
• Add Snowflake Artic Embed 2.0 model by @s-kostyaev in #125
• docs: update README.org by @eltociear in #126
• Add gemini-2.0-flash-thinking-exp, fix output selection by @ahyatt in #127
• Fix missing capabilities in Gemini thinking model by @ahyatt in #128
• Fix integration test false positives by @ahyatt in #129

New Contributors

• @jsntn made their first contribution in #118
• @eltociear made their first contribution in #126

Full Changelog: 0.19.1...0.20.0

Fix Open AI's model context sizes

What's Changed

• Reduce Open AI's context length; most users will have smaller ctxt by @ahyatt in #116

Full Changelog: 0.19.0...0.19.1

json-mode and functions as keys

What's Changed

• Add BGE-M3 multilingual embedding model by @s-kostyaev in #109
• Add JSON mode for the providers that support it by @ahyatt in #112
• Add support for auth-source secret functions to :key argument by @minad in #111
• Extend key function to other models, add to README by @ahyatt in #114

New Contributors

• @minad made their first contribution in #111

Full Changelog: 0.18.1...0.19.0

FIx for llm-batch-embeddings-async

What's Changed

• Fix for ollama name and capabilities for embedding only models by @ahyatt in #106
• Fix extra argument in llm-batch-embeddings-async by @ahyatt in #107

Full Changelog: 0.18.0...0.18.1

0.18.0

What's Changed

• Add Azure support by @ahyatt in #96
• Add multimodal support for Openai, Gemini, and Ollama by @awswan in #88
• Add batch embeddings capability, implement for OpenAI and Ollama by @ahyatt in #93
• Add ability to fill prompt variable backwards by @ahyatt in #82
• Add new ollama models to the list of function calling models by @ahyatt in #84
• Centralizing model information in new llm-models.el by @ahyatt in #85
• Unvendor plz-event-source and plz-media-type by @leotaku in #87
• Ignore tests and utilities for elpa by @ahyatt in #89
• Update Claude Sonnet version by @ahyatt in #90
• Fix issue with empty function call args for Open AI by @ahyatt in #99
• Handle plz-error-message error type by @leotaku in #95

New Contributors

• @leotaku made their first contribution in #87
• @awswan made their first contribution in #88

Full Changelog: 0.17.4...0.18.0

Add llm-prompt-default-max-tokens, open AI token limit fixes, parallel tool use fixes

What's Changed

• Fix breakage with Open AI's llm-chat-token-limit by @ahyatt in #77
• Fix Vertex and Open AI's parallel call tool use by @ahyatt in #78
• Add variable llm-prompt-default-max-tokens by @ahyatt in #79
• Fix how we look for ollama models in integration tests by @ahyatt in #80

Full Changelog: 0.17.3...0.17.4

More efficient, streaming text insertion, function calling conversational fixes

What's Changed

• Make streaming not repeatedly insert the same thing by @ultronozm in #72
• Fix error with ollama function results by @ultronozm in #74
• Fix bug involving multiple function calls with Claude by @ultronozm in #73
• Remove the debug logs output on streaming, to reduce log volume by @ahyatt in #75

Full Changelog: 0.17.2...0.17.3