Skip to main content
llm.kiwi provides access to a curated selection of high-performance AI models. Model availability depends on your subscription tier.

Free Tier

The Free tier includes access to a single routing model:

default

Intelligent model router that selects the best model for your request automatically.
Model IDDescription
defaultAuto-routing model — balances quality and speed for each request
Free tier users must use model="default". This model intelligently routes your requests to appropriate underlying models.

Pro Tier Models

Pro tier unlocks direct access to specific high-performance models:

Chat & Reasoning Models

Model IDProviderDescription
gpt-oss-20bOpenAI OSS20B parameter open-source GPT variant
gpt-oss:20bOpenAI OSSAlternative syntax for gpt-oss-20b
mistral-small-3.1-24b-instruct-2503Mistral24B instruction-tuned model
meta-llama/Meta-Llama-3.1-8B-Instruct-TurboMetaLlama 3.1 8B optimized for speed
deepseek-v3.1:671b-terminusDeepSeek671B MoE model, excellent for coding
gemini-2.5-flash-liteGoogleUltra-fast reasoning model
GLM-4.6V-FlashZhipuFast Chinese/English bilingual
bidaraBidaraBiomimicry design assistant

Code Models

Model IDProviderDescription
codestral-2405MistralCode generation (May 2024 release)
codestral-2501MistralCode generation (Jan 2025 release)
ministral-8b-2512MistralCompact 8B model for fast responses

Image & Audio Models

Model IDProviderTierDescription
fluxFluxProHigh-quality image generation
whisper-1OpenAIProSpeech-to-text transcription

Choosing a Model

Not sure which model to use? Start with default — it automatically routes to the best model for your use case.
Use CaseRecommended Model
General chatdefault or gpt-oss-20b
Complex codingdeepseek-v3.1:671b-terminus or codestral-2501
Fast responsesgemini-2.5-flash-lite or ministral-8b-2512
Chinese/EnglishGLM-4.6V-Flash

Upgrade to Pro

Unlock all Pro models and higher rate limits.