Providers

Elyxir connects models from multiple providers. This guide describes provider families and their broader capabilities; it does not guarantee that every provider feature or direct connection is currently enabled in Elyxir. Check the supported model catalog, GET /v1/models, and the API overview for callable model IDs and endpoints.

Provider-native names shown below are background reference only. For Elyxir requests, use the exact model IDs returned by the API, without adding a provider prefix. Media and embedding capabilities are not a promise of corresponding public API endpoints.

Provider Overview

OpenAI

Website: openai.com

Capabilities

• Chat completions
• Vision (GPT-4.1, GPT-4o)
• Embeddings
• Image generation (gpt-image-2, gpt-image-1.5)
• Audio transcription (Whisper)

Strengths

• Industry-leading quality
• Fast response times
• Wide ecosystem support
• Strong instruction following

Provider-native names (not Elyxir request IDs)

openai/gpt-5.5
openai/gpt-5.4
openai/gpt-5.2
openai/gpt-5-mini
openai/gpt-4.1
openai/gpt-4.1-mini
openai/gpt-4o

Anthropic

Website: anthropic.com

Capabilities

• Chat completions
• Vision (Claude 4+)
• Very long context (200K tokens)
• Computer use
• Extended thinking
• Agentic coding

Strengths

• Exceptional coding abilities
• Detailed, thoughtful responses
• Strong safety measures
• Long context handling
• Best-in-class for software development

Provider-native names (not Elyxir request IDs)

anthropic/claude-opus-4.8
anthropic/claude-opus-4.7
anthropic/claude-sonnet-4.6
anthropic/claude-opus-4.6
anthropic/claude-sonnet-4.5
anthropic/claude-opus-4.5
anthropic/claude-haiku-4.5

Google

Website: ai.google.dev

Capabilities

• Chat completions
• Vision
• Multimodal (text, image, audio, video)
• Extremely long context (1M tokens)
• Code execution
• Image generation (Imagen 4)
• Video generation (Veo 3)

Strengths

• Massive context windows
• Native multimodal support
• Good price-performance
• Fast inference
• Google Search grounding

Provider-native names (not Elyxir request IDs)

gemini/gemini-3.5-flash
gemini/gemini-3.1-pro-preview
gemini/gemini-3-pro-preview
gemini/gemini-3-flash

Meta

Website: llama.meta.com

Capabilities

• Chat completions
• Code generation
• Multilingual
• Vision (Llama 4)

Strengths

• Open weights (can self-host)
• No usage restrictions
• Strong performance
• Active community
• Multiple size options

Provider-native names (not Elyxir request IDs)

meta/llama4-scout
meta/llama4-maverick
meta/llama-3.3-70b

Mistral AI

Website: mistral.ai

Capabilities

• Chat completions
• Code generation (Codestral)
• Function calling
• Multilingual

Strengths

• European data sovereignty
• Efficient architecture
• Strong code generation
• Competitive pricing
• Open weight options

Provider-native names (not Elyxir request IDs)

mistral/mistral-large-3
mistral/mistral-medium-3.1
mistral/mistral-small-4
mistral/mistral-small-3.2
mistral/codestral-22b

DeepSeek

Website: deepseek.com

Capabilities

• Chat completions
• Code generation
• Advanced reasoning
• Math and science

Strengths

• State-of-the-art reasoning
• Excellent at math problems
• Strong coding abilities
• Very cost-effective
• Open weights available

Provider-native names (not Elyxir request IDs)

deepseek-v4-pro
deepseek-v4-flash
deepseek-v3.2
deepseek-v3.1

Cohere

Website: cohere.com

Capabilities

• Chat with RAG
• Enterprise-grade reliability
• Function calling
• Vision (Command A+)

Strengths

• Excellent for enterprise
• Strong RAG support
• Reliable performance

Provider-native names (not Elyxir request IDs)

command-a-plus
command-a

Perplexity

Website: perplexity.ai

Capabilities

• Chat with web search
• Real-time information
• Source citations
• Deep multi-step research
• Reasoning with search grounding

Strengths

• Native web search
• Source citations
• Current information
• Research tasks

Provider-native names (not Elyxir request IDs)

sonar-deep-research
sonar-reasoning-pro
perplexity-sonar-large
perplexity-sonar

xAI

Website: x.ai

Capabilities

• Chat completions
• Real-time information
• Deep reasoning
• Agentic coding (Grok Build)

Strengths

• Current events awareness
• Large context window (up to 1M)
• Strong reasoning
• Cost-efficient frontier tier (grok-4.3)

Provider-native names (not Elyxir request IDs)

xai/grok-4.3
xai/grok-4
xai/grok-build-0.1

Fireworks AI

Website: fireworks.ai

Capabilities

• Chat completions
• Ultra-fast inference
• Function calling
• Vision models

Strengths

• Very fast inference
• Wide model selection including Qwen series
• Competitive pricing
• Enterprise features

Available Models

Fireworks AI hosts a variety of open models, including:

• Qwen series: qwen3.5-397b-a17b, qwen3-coder-next, qwen3.7-max
• Kimi series: kimi-k2.6, kimi-k2, kimi-k2-thinking
• DeepSeek: deepseek-v4-pro, deepseek-v4-flash, deepseek-v3.2, deepseek-v3.1
• Llama 4: llama4-scout, llama4-maverick
• GLM: glm-4.7
• MiniMax: minimax-m2.7

Provider-native names (not Elyxir request IDs)

fireworks/kimi-k2.6
fireworks/deepseek-v4-pro
fireworks/deepseek-v4-flash
fireworks/deepseek-v3.2
fireworks/llama4-scout

ElevenLabs

Website: elevenlabs.io

Capabilities

• Text-to-speech (ElevenLabs v3)
• Voice cloning
• Speech-to-text transcription (Scribe)
• Audio generation

Strengths

• Natural-sounding voices
• Multiple languages
• Custom voices
• Fast generation
• High-accuracy transcription

Provider-native names (not Elyxir request IDs)

elevenlabs/eleven-v3
elevenlabs/scribe-v2
elevenlabs/scribe-v1
elevenlabs/scribe-v1-experimental

Choosing a Provider

By Use Case

By Priority

Related

• Supported Models - Model list
• Model Hub - Browse models
• Comparison - Compare options