Skip to content

Provider & Models List

Tarsk supports a large and growing number of model providers. Each provider requires its own API key (or token for some). You can enable multiple providers simultaneously and switch between models per thread.

These providers host their own models. Bring an API key from the provider’s website.

Models: Claude Opus 4.6, Claude Sonnet 4.6, Claude Sonnet 4.5, Claude Haiku 4.5 Get a key: console.anthropic.com

Claude models excel at coding, reasoning, and following complex instructions.

Models: GPT-5, GPT-4.1, GPT-4.1 mini, o3, o3-pro, o4-mini, Codex series Get a key: platform.openai.com

Models: Gemini 3.1 Pro Preview, Gemini 3 Pro Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.5 Flash Lite Get a key: aistudio.google.com

Gemini 2.5 Flash is a fast, cost-effective option with a large context window.

Models: Grok 4, Grok 4.1 Fast, Grok 3, Grok Code Fast 1 Get a key: console.x.ai

Models: DeepSeek Reasoner (R1), DeepSeek Chat (V3) Get a key: platform.deepseek.com

DeepSeek R1 is a strong reasoning model. DeepSeek V3 is competitive with frontier models for coding tasks.

Models: Devstral Medium, Magistral Medium, Mistral Large 3, Mistral Small 3.2, Codestral Get a key: console.mistral.ai

Devstral is Mistral’s code-focused model family. Codestral targets code completion and generation.

Models: Sonar Pro, Sonar Reasoning Pro, Sonar Deep Research Get a key: perplexity.ai

Sonar models are augmented with live web search.

Models: Command A, Command A Reasoning, Command R+, Command R Get a key: dashboard.cohere.com

Models: Llama 4 Scout 17B, Llama 4 Maverick 17B, Kimi K2 Instruct, GPT OSS 120B, Qwen3 32B Get a key: console.groq.com

Groq provides very fast inference via custom hardware.

Models: Qwen 3 235B Instruct, GPT OSS 120B, Llama 3.1 8B, Z.AI GLM-4.7 Get a key: cloud.cerebras.ai

Models: Kimi K2.5, Kimi K2 Thinking, Kimi K2 Turbo Variants: moonshotai (international), moonshotai-cn (China endpoint) Get a key: platform.moonshot.ai

Models: MiniMax-M2.5, MiniMax-M2.1, MiniMax-M2 Variants: minimax (international), minimax-cn (China endpoint) Get a key: platform.minimax.io

Models: GLM-5, GLM-4.7, GLM-4.6, GLM-4.5 Variants: zhipuai, zai Get a key: bigmodel.cn

Models: Qwen3 Coder 480B, Qwen3.5 397B, Qwen3 235B, Qwen Max, Qwen3 Coder Flash Get a key: dashscope.aliyuncs.com

Models: Nemotron series, Phi series, DeepSeek V3.x, Kimi K2, Qwen3 Coder 480B, GLM-4.7 Get a key: build.nvidia.com

Models: GLM-5, GLM-4.7, MiMo-V2-Flash, MiniMax-M2.5, DeepSeek-R1, Kimi K2, Qwen3 Coder 480B Get a key: huggingface.co/settings/tokens


Aggregators route requests to multiple underlying models through a single API key. They are useful for accessing many models without managing separate keys.

Models: Hundreds of models from Anthropic, OpenAI, Google, xAI, Meta, Mistral, and more Get a key: openrouter.ai Credits: Balance visible in Settings

OpenRouter is the most comprehensive aggregator. A single key accesses almost every major model. Some models are free; paid models are charged per-token.

Models: Claude, GPT-5, Gemini, DeepSeek, Kimi, GLM, Qwen via a single API Get a key: aihubmix.com Credits: Balance visible in Settings

Models: Qwen, DeepSeek, Kimi, GLM, MiniMax, Llama Variants: siliconflow (international), siliconflow-cn (China endpoint) Get a key: cloud.siliconflow.com

Models: DeepSeek, Kimi, Qwen, GLM Get a key: platform.iflow.cn

Models: Qwen, GLM, DeepSeek Get a key: modelscope.cn

Models: GLM, MiniMax, DeepSeek, Kimi, Qwen, Llama, GPT OSS Get a key: api.together.ai

Models: Kimi, GLM, DeepSeek, MiniMax, GPT OSS Get a key: fireworks.ai

Models: GLM, MiniMax, DeepSeek, Kimi, Llama, Qwen, Claude, GPT OSS Get a key: deepinfra.com

Models: GLM, MiniMax, DeepSeek, Kimi, ERNIE, Qwen, Llama, GPT OSS Get a key: novita.ai

Models: GLM, MiniMax, DeepSeek, Kimi, Qwen, Llama, Gemma, Hermes, INTELLECT Get a key: tokenfactory.nebius.com

Models: Multi-provider aggregator — DeepSeek, Kimi, GLM, Qwen, Grok, GPT, Claude, Gemini, MiniMax Get a key: zenmux.ai

Models: Multi-provider aggregator — Claude, GPT, Gemini, Kimi, GLM, MiniMax, Qwen, Grok Get a key: opencode.ai

Models: Claude, GPT, Gemini, Grok, GLM, Kimi, MiniMax (via Poe subscription) Get a key: poe.com


These providers give you access to foundation models through your existing cloud account.

Models: GPT-5, GPT-4.1, o3, o4-mini, Grok 4, Claude, Mistral, Phi, DeepSeek, Kimi, Llama Auth: AZURE_RESOURCE_NAME + AZURE_API_KEY Docs: learn.microsoft.com/en-us/azure/ai-services/openai

Models: Gemini 3 Pro Preview, Gemini 2.5 Pro, Gemini 2.5 Flash, DeepSeek, GLM, Qwen, Llama, GPT OSS Auth: GOOGLE_VERTEX_PROJECT + GOOGLE_VERTEX_LOCATION + GOOGLE_APPLICATION_CREDENTIALS Docs: cloud.google.com/vertex-ai

Models: Claude, Llama, Mistral, DeepSeek, Kimi, Qwen, MiniMax, GLM, NVIDIA Nemotron, Nova Auth: AWS_ACCESS_KEY_ID + AWS_SECRET_ACCESS_KEY + AWS_REGION Docs: docs.aws.amazon.com/bedrock


These providers use a personal access token rather than a separate API key.

Models: Claude Opus 4.6, Claude Sonnet 4.6, GPT-5, Gemini 3 Pro Preview, Grok Code Fast 1 Auth: Connect via your GitHub account — no separate API key needed if you have a Copilot subscription

Models: Phi-4, AI21 Jamba, Llama, Mistral and more (preview access) Auth: GITHUB_TOKEN Docs: github.com/marketplace/models

Models: Claude Opus 4.6, Claude Sonnet 4.5, Claude Haiku 4.5, GPT-5, GPT-5 Mini Auth: GITLAB_TOKEN Docs: docs.gitlab.com/ee/user/gitlab_duo


Run models on your own hardware.

Models: Local models you download (Qwen3, GPT OSS, and more) Auth: LMSTUDIO_API_KEY Docs: lmstudio.ai

Models: Cloud-hosted Ollama models — Kimi K2, GLM, Qwen3, Gemma, DeepSeek, MiniMax Auth: OLLAMA_API_KEY Docs: docs.ollama.com/cloud


ProviderModelsNotes
zai-coding-planGLM-5, GLM-4.7-FlashX, GLM-4.5Z.AI planning-focused variants
zhipuai-coding-planGLM-5, GLM-4.7, GLM-4.6, GLM-4.5Zhipu AI coding + planning split
kimi-for-codingKimi K2.5, Kimi K2 ThinkingKimi specialised coding mode

Next: Configure providers and enable models →