Add image upload support for Dola provider

Strengthen tool prompt instructions with examples

Add tool/function calling support for all providers

Add Dola provider (dola-fast / dola-pro) with reverse-engineered chat/completion endpoint

Fix missing imports: server, logger, http, cookieJar come from @exocore/multi, not src/index.js

Fix missing cookieJar import — removed from dead import, replaced with clearCookiesFile

Add image support for /v1 endpoints: fix extractImageUrls for Anthropic format, fix Qwen image_url handling, add Gemini image upload

Add DeepSeek Instant/Expert/Vision modes + modelType + image support in /v1

Add Meta AI fast (think_fast) and thinking (think_hard) modes + model variants

Fix Meta AI: fix token extraction regex, improve response parsing, add auth error handling, add meta models to ANTHRO_MODEL_MAP

Add fast/thinking mode variants for all Qwen models + meta module

Update Exocore Gateway

Revert ANTHRO_MODEL_MAP and kimi.json to original

Add NVIDIA Kimi K2.6 API for Claude Code tool support

Switch Claude Code backend from DeepSeek to Kimi

Switch Claude Code backend from Qwen to DeepSeek

Switch to Qwen3.5-Flash for faster Claude Code responses

Route all Claude model names to Qwen3.7-Max for Claude Code support

Add GPT and Grok via Puter driver + V1 routes and HTML docs

Add Claude integration via Puter driver API

Add tool-discipline instruction: models should not waste tool calls on greetings

Stronger system prompt override for text-only models (Gemini/DeepSeek/Kimi)

Fix system prompt array→text for all providers + embed tool descriptions in text fallback

Add OpenAI-compatible tool passthrough for DeepSeek and Kimi (optional apiKey field)

Fix system prompt array→string conversion in convertAnthropicToOpenAI

Fix tool format conversion (Anthropic→OpenAI), strip <details> tags, add debug logging

Add generic callOpenAIV1 for any OpenAI-compatible provider. All models get tool support when API key is configured.

Add full tool calling support for Qwen via askQwenV1 (Anthropic-to-OpenAI conversion)

Add qwen3.7-max support, set as default model

Fix model routing: check V1_MODEL_MAP before defaulting to gemini

Update server.js

Update server.js

Update server.js

Move config JSONs (deepseek, qwen, kimi) to data/ folder

Update Gemini API to support new AI models and improve functionality

Update server.js

Upload 20 files