Skip to main content
LLM.kiwi LogoLLM.kiwi
Model Catalog v2.5

Every Major LLM.
One Simple Endpoint.

Access the world's most advanced language, vision, and image models through LLM.kiwi. No per-token pricing, no vendor lock-in.

LLM.kiwi Core

LLM.kiwi Core

Official auto-routing models for maximum performance.

Default (Auto)Free
default
CAPABILITY
Smart Routing
Fast (Speed)Free
fast
CAPABILITY
Ultra Latency
Pro (High Cap)Pro
pro
CAPABILITY
Max Reasoning
OpenAI

OpenAI

State-of-the-art models for general and technical tasks.

GPT-5 MiniPro
gpt-5-mini
CAPABILITY
Advanced Reasoning
GPT-4.1 NanoPro
gpt-4.1-nano-2025-04-14
CAPABILITY
Fast Reasoning
WhisperPro
whisper
CAPABILITY
Speech to Text
Google Gemini

Google Gemini

Large context and search-grounded responses.

Gemini SearchPro
gemini-search
CAPABILITY
Web Grounding
Gemini 2.5 Flash LiteFree
gemini-2.5-flash-lite
CAPABILITY
High Efficiency
DeepSeek

DeepSeek

Intelligent models optimized for coding and math.

DeepSeek V3.1Pro
deepseek-v3.1
CAPABILITY
Coding Specialist
Mistral AI

Mistral AI

European high-performance open models.

Codestral 2501Pro
codestral-2501
CAPABILITY
Code Generation
Mistral Small 3.1Pro
mistral-small-3.1-24b-instruct-2503
CAPABILITY
Balanced logic
Ministral 8BPro
ministral-8b-2512
CAPABILITY
Edge Optimized
Meta Llama

Meta Llama

Popular open source models with massive throughput.

Llama 3.1 8B TurboPro
meta-llama/Meta-Llama-3.1-8B-Instruct-Turbo
CAPABILITY
Lightning Speed
Specialized

Specialized

Unique models for images and tasks.

FLUXPro
flux
CAPABILITY
Image Gen
GLM 4.5 FlashPro
glm-4.5-flash
CAPABILITY
Bilingual Polish
BIDARAFree
bidara
CAPABILITY
Biomimicry Design

Zero Latency Path

Optimized routing to ensure your requests hit the fastest available model node.

Unbiased Selection

We don't favor any provider. You get the raw, untouched power of every model.

Global Availability

Deployed strategically across 12+ regions to minimize geo-latency.

The Future of AI is Unified

Start building with the world's best models today. No per-token limitations. No vendor lock-in.

Start Building