Best Free AI Models: Compare Today's LLM Rankings
Compare the best free models for coding, writing, reasoning and multimodal work before choosing an API provider.
Last updated: 2026-08-23
How to choose a free AI model
Start with the task: use text rankings for writing and Q&A, code rankings for programming, and multimodal models for images or documents. Then compare context window, rate limits and provider availability. Test shortlisted models with your own prompts before relying on them in production.
Available free models
| Platform | Model | Text rank | Code rank | Type | Context | Visit platform |
|---|---|---|---|---|---|---|
| OpenRouter |
nvidia/nemotron-3-nano-omni-30b-a3b-reasoning:free
nvidia
|
OR #10 |
- |
reasoning | 256000 | Try model |
| OpenRouter |
google/gemma-4-31b-it:free
Google
|
OR #18 |
- |
text | 262144 | Try model |
| OpenRouter |
google/gemma-4-26b-a4b-it:free
Google
|
OR #15 |
- |
text | 262144 | Try model |
| OpenRouter |
nvidia/nemotron-3-super-120b-a12b:free
nvidia
|
OR #5#187 1360 |
- |
text | 262144 | Try model |
| OpenRouter |
nvidia/nemotron-3-nano-30b-a3b:free
nvidia
|
OR #9 |
- |
text | 256000 | Try model |
| OpenRouter |
nvidia/nemotron-nano-9b-v2:free
nvidia
|
OR #13 |
- |
text | 128000 | Try model |
| OpenRouter |
thinkingmachines/inkling:free
thinkingmachines
|
OR #12 |
- |
reasoning | 262144 | Try model |
| OpenRouter |
thinkingmachines/inkling-small:free
thinkingmachines
|
OR #17 |
- |
reasoning | 262144 | Try model |
| OpenRouter |
google/lyria-3-clip-preview
Google
|
OR #21 |
- |
text | 1048576 | Try model |
| OpenRouter |
google/lyria-3-pro-preview
Google
|
OR #20 |
- |
text | 1048576 | Try model |
| OpenRouter |
cohere/north-mini-code:free
cohere
|
OR #6 |
- |
text | 256000 | Try model |
| OpenRouter |
z-ai/glm-5.2:free
Zhipu AI
|
OR #11 |
- |
text | 256000 | Try model |
| OpenRouter |
dots-studio/dots-3-note-preview:free
dots-studio
|
OR #8 |
- |
text | 512000 | Try model |
| OpenRouter |
poolside/laguna-xs-2.1:free
poolside
|
OR #7 |
- |
text | 262144 | Try model |
| OpenRouter |
stealth/ox-alpha
stealth
|
OR #1 |
- |
text | 1048576 | Try model |
| OpenRouter |
nvidia/nemotron-3.5-lightning:free
nvidia
|
OR #4 |
- |
text | 1000000 | Try model |
| OpenRouter |
poolside/laguna-s-2.1:free
poolside
|
OR #3 |
- |
text | 262144 | Try model |
| OpenRouter |
nvidia/nemotron-3-ultra-550b-a55b:free
nvidia
|
OR #2 |
- |
text | 1000000 | Try model |
| OpenRouter |
openrouter/free
openrouter
|
OR #22 |
- |
text | 200000 | Try model |
| OpenRouter |
nvidia/nemotron-3.5-content-safety:free
nvidia
|
OR #19 |
- |
text | 128000 | Try model |
| OpenRouter |
liquid/lfm-2.5-2.6b:free
liquid
|
OR #16 |
- |
text | 65536 | Try model |
| OpenRouter |
nvidia/nemotron-nano-12b-v2-vl:free
nvidia
|
OR #14 |
- |
multimodal | 128000 | Try model |
| SiliconFlow |
THUDM/GLM-Z1-9B-0414
THUDM
|
- |
- |
reasoning | 131072 | Try model |
| SiliconFlow |
Qwen/Qwen3.5-4B
Alibaba/Qwen
|
- |
- |
text | 262144 | Try model |
| SiliconFlow |
deepseek-ai/DeepSeek-R1-0528-Qwen3-8B
DeepSeek
|
- |
- |
reasoning | 131072 | Try model |
| SiliconFlow |
Qwen/Qwen3-8B
Alibaba/Qwen
|
- |
- |
text | 131072 | Try model |
| SiliconFlow |
THUDM/GLM-4-9B-0414
THUDM
|
- |
- |
text | 32768 | Try model |
| SiliconFlow |
deepseek-ai/DeepSeek-OCR
DeepSeek
|
- |
- |
text | 8192 | Try model |
| SiliconFlow |
Qwen/Qwen2.5-7B-Instruct
Alibaba/Qwen
|
- |
- |
text | 32768 | Try model |
| SiliconFlow |
tencent/Hunyuan-MT-7B
tencent
|
- |
- |
text | 32768 | Try model |
| SiliconFlow |
TeleAI/TeleSpeechASR
TeleAI
|
- |
- |
text | - | Try model |
| SiliconFlow |
Kwai-Kolors/Kolors
Kwai-Kolors
|
- |
- |
text | - | Try model |
| SiliconFlow |
BAAI/bge-m3
BAAI
|
- |
- |
text | - | Try model |
| SiliconFlow |
BAAI/bge-large-en-v1.5
BAAI
|
- |
- |
text | - | Try model |
| SiliconFlow |
BAAI/bge-large-zh-v1.5
BAAI
|
- |
- |
text | - | Try model |
| SiliconFlow |
BAAI/bge-reranker-v2-m3
BAAI
|
- |
- |
text | - | Try model |
| SiliconFlow |
FunAudioLLM/SenseVoiceSmall
FunAudioLLM
|
- |
- |
text | - | Try model |
| SiliconFlow |
Qwen/Qwen3-ASR-1.7B
Alibaba/Qwen
|
- |
- |
text | - | Try model |
| SiliconFlow |
PaddlePaddle/PaddleOCR-VL-1.5
PaddlePaddle
|
- |
- |
multimodal | - | Try model |
| 百度飞桨 |
paddleocr-vl-1.5
Baidu
|
- |
- |
multimodal | 8192 | Try model |
| Hugging Face |
Qwen/Qwen2.5-0.5B-Instruct
Alibaba/Qwen
|
- |
- |
text | 32768 | Try model |
| Hugging Face |
openai/whisper-base
OpenAI
|
- |
- |
text | - | Try model |
| Hugging Face |
hexgrad/Kokoro-82M
hexgrad
|
- |
- |
text | - | Try model |
About these rankings
This page lists currently available free models and their benchmark-based rankings. OpenRouter's weekly token-usage order is the primary ranking signal, with AI Arena used when no OpenRouter ranking is available. Availability, limits and model versions can change by provider. Free tiers are useful for evaluation and smaller workloads; review each provider's terms before production use.