Free AI models for AI agents
Updated:
Models with tool calling, reasoning, and long context for autonomous and assisted agents: planning, tool use, and multi-step work without paying for tokens.
Read more
The table favors strong instruct and reasoning routes. Before running an agent, confirm structured output, function calling, and how the provider refreshes its free quota.
110 models found
| № | Model | Power | Provider | Parameters | Context | Limit | |
|---|---|---|---|---|---|---|---|
| 1 |
Qwen: Qwen3.8 27B (free)
|
⚖️ | OpenRouter | 27B | 262K | Free tier | Open |
| 2 |
NVIDIA: Nemotron 3.5 Lightning (free)
|
⚖️ | OpenRouter | ? | 1.0M | Free tier | Open |
| 3 |
NVIDIA: Nemotron 3.5 Content Safety (free)
|
⚖️ | OpenRouter | ? | 128K | Free tier | Open |
| 4 |
NVIDIA: Nemotron 3 Ultra (free)
|
🚀 | OpenRouter | 550B | 1.0M | Free tier | Open |
| 5 |
NVIDIA: Nemotron 3 Nano Omni (free)
|
⚖️ | OpenRouter | 30B | 256K | Free tier | Open |
| 6 |
NVIDIA: Nemotron 3 Super (free)
|
🚀 | OpenRouter | 120B | 262K | Free tier | Open |
| 7 |
Qwen3.8 27b
|
⚖️ | Groq | 27B | 131K | 14,400 req/day | Open |
| 8 |
Gemini 2.0 Flash
|
⚖️ | Google Gemini | ? | 1.0M | 1,500 req/day | Open |
| 9 |
Gemini 2.0 Flash Lite
|
⚖️ | Google Gemini | ? | 1.0M | 1,500 req/day | Open |
| 10 |
Gemini 1.5 Flash
|
⚖️ | Google Gemini | ? | 1.0M | 1,500 req/day | Open |
| 11 |
Gemini 1.5 Pro
|
🚀 | Google Gemini | ? | 2.1M | 50 req/day | Open |
| 12 |
Mistral Small
|
⚖️ | Mistral AI | ? | 33K | 1 req/min · 500 tok/min | Open |
| 13 |
Mistral Nemo
|
⚖️ | Mistral AI | ? | 131K | 1 req/min · 500 tok/min | Open |
| 14 |
Mistral Embed
|
⚖️ | Mistral AI | ? | 8K | 1 req/min · 500 tok/min | Open |
| 15 |
nim/meta/llama-3.2-90b-vision-instruct
|
🚀 | Together AI | 90B | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 16 |
nim/mistralai/mixtral-8x22b-instruct-v01
|
⚖️ | Together AI | ? | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 17 |
nim/meta/llama-3.3-70b-instruct
|
🚀 | Together AI | 70B | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 18 |
nim/nvidia/llama-3.1-nemotron-70b-instruct
|
🚀 | Together AI | 70B | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 19 |
nim/meta/llama-3.1-70b-instruct
|
🚀 | Together AI | 70B | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 20 |
nim/nv-mistralai/mistral-nemo-12b-instruct
|
⚖️ | Together AI | 12B | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 21 |
nim/mistralai/mixtral-8x7b-instruct-v01
|
⚖️ | Together AI | ? | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 22 |
nim/nvidia/llama-3.3-nemotron-super-49b-v1
|
⚖️ | Together AI | 49B | 16K | Fair Use (По мере нагрузки серверов) | Open |
| 23 |
Cogito V1 Preview Qwen 32B
|
⚖️ | Together AI | 32B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 24 |
Cogito V1 Preview Qwen 14B
|
⚖️ | Together AI | 14B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 25 |
Cogito V1 Preview Llama 70B
|
🚀 | Together AI | 70B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 26 |
Cogito V1 Preview Llama 70B Turbo
|
🚀 | Together AI | 70B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 27 |
Meta Llama 3.3 70B Instruct
|
🚀 | Together AI | 70B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 28 |
Qwen2.5 32B
|
⚖️ | Together AI | 32B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 29 |
Qwen2.5 72B
|
🚀 | Together AI | 72B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 30 |
Qwen2.5 14B
|
⚖️ | Together AI | 14B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 31 |
Llama 3.1 70B
|
🚀 | Together AI | 70B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 32 |
Qwen2.5 32B Instruct
|
⚖️ | Together AI | 32B | 33K | Fair Use (По мере нагрузки серверов) | Open |
| 33 |
Llama 3.1 405B
|
🚀 | Together AI | 405B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 34 |
Devstral Small 2505
|
⚖️ | Together AI | ? | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 35 |
Mixtral 8X22b Instruct V0.1
|
⚖️ | Together AI | ? | 66K | Fair Use (По мере нагрузки серверов) | Open |
| 36 |
Qwen3 14B
|
⚖️ | Together AI | 14B | 2K | Fair Use (По мере нагрузки серверов) | Open |
| 37 |
Qwen3 30B A3b
|
⚖️ | Together AI | 30B | 41K | Fair Use (По мере нагрузки серверов) | Open |
| 38 |
Llama 4 Scout (17Bx16E)
|
⚖️ | Together AI | 17B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 39 |
Magistral Small 2506
|
⚖️ | Together AI | ? | 41K | Fair Use (По мере нагрузки серверов) | Open |
| 40 |
Qwen3 32B
|
⚖️ | Together AI | 32B | 41K | Fair Use (По мере нагрузки серверов) | Open |
| 41 |
Qwen3 235B A22b Instruct 2507 Fp8
|
🚀 | Together AI | 235B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 42 |
Qwen3 Coder 30B A3b Instruct
|
⚖️ | Together AI | 30B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 43 |
Qwen3 Next 80B A3b Instruct Fp8
|
🚀 | Together AI | 80B | ? | Fair Use (По мере нагрузки серверов) | Open |
| 44 |
Qwen3-VL-235B-A22B-Instruct-FP8
|
🚀 | Together AI | 235B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 45 |
Nvidia Nemotron 3 Nano 30B A3b Bf16
|
⚖️ | Together AI | 30B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 46 |
Qwen3.5 35B A3b
|
⚖️ | Together AI | 35B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 47 |
Nvidia Nemotron 3 Super 120B A12b Fp8
|
🚀 | Together AI | 120B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 48 |
Deepseek OCR 2
|
⚖️ | Together AI | ? | 8K | Fair Use (По мере нагрузки серверов) | Open |
| 49 |
Qwen3.5 122B A10b Fp8
|
🚀 | Together AI | 122B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 50 |
Qwen3 30B A3B Instruct 2507 Lora
|
⚖️ | Together AI | 30B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 51 |
Nvidia Nemotron 3 Super 120B A12b Bf16
|
🚀 | Together AI | 120B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 52 |
Qwen3.6 35B A3b Fp8
|
⚖️ | Together AI | 35B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 53 |
Nemotron 3 Nano Omni 30B A3b Reasoning Fp8
|
⚖️ | Together AI | 30B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 54 |
Llama 3.3 70B Instruct FP8 Lora
|
🚀 | Together AI | 70B | 131K | Fair Use (По мере нагрузки серверов) | Open |
| 55 |
Mixtral 8x7B Instruct V0.1 FP8 Lora
|
⚖️ | Together AI | ? | 33K | Fair Use (По мере нагрузки серверов) | Open |
| 56 |
Llama 4 Scout 17B 16E Instruct Fp8 Lora
|
⚖️ | Together AI | 17B | 10.5M | Fair Use (По мере нагрузки серверов) | Open |
| 57 |
Llama 4 Maverick 17B 128E Instruct Nvfp4
|
⚖️ | Together AI | 17B | 1.0M | Fair Use (По мере нагрузки серверов) | Open |
| 58 |
Qwen3.5 35B A3B Lora
|
⚖️ | Together AI | 35B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 59 |
Qwen3.6 35B A3B Lora
|
⚖️ | Together AI | 35B | 262K | Fair Use (По мере нагрузки серверов) | Open |
| 60 |
Qwen 3.8 27b
|
⚖️ | Cerebras | 27B | ? | 14,400 req/day · 30 req/min | Open |
| 61 |
Qwen 1.5 14B Chat
|
⚖️ | Cloudflare | 14B | 33K | 10,000 req/day | Open |
| 62 |
DeepSeek: DeepSeek V4 Flash (Free)
|
🚀 | OrcaRouter | ? | ? | Free tier | Open |
| 63 |
Google: Nano Banana Pro (Gemini 3 Pro Image Preview)
|
🚀 | OrcaRouter | ? | 66K | Free tier | Open |
| 64 |
Google: Nano Banana 2 (Gemini 3.1 Flash Image Preview)
|
⚖️ | OrcaRouter | ? | 66K | Free tier | Open |
| 65 | Codellama 70b | 🚀 | NVIDIA Build | 70B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 66 |
Codestral 22b Instruct V0.1
|
⚖️ | NVIDIA Build | 22B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 67 |
Deepseek V4.1 Flash
|
🚀 | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 68 | Granite 34b Code Instruct | ⚖️ | NVIDIA Build | 34B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 69 |
Jamba 1.5 Large Instruct
|
🚀 | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 70 |
LLAMA 3.1 Nemotron 51b Instruct
|
⚖️ | NVIDIA Build | 51B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 71 |
LLAMA 3.1 Nemotron 70b Instruct
|
🚀 | NVIDIA Build | 70B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 72 |
LLAMA 3.1 Nemotron Ultra 253b V1
|
🚀 | NVIDIA Build | 253B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 73 |
LLAMA 3.2 90b Vision Instruct
|
🚀 | NVIDIA Build | 90B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 74 | Llama2 70b | 🚀 | NVIDIA Build | 70B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 75 |
Llama3 Chatqa 1.5 70b
|
🚀 | NVIDIA Build | 70B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 76 |
Mistral Large
|
🚀 | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 77 |
Mistral Large 2 Instruct
|
🚀 | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 78 |
Mistral Nemo 12b Instruct
|
⚖️ | NVIDIA Build | 12B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 79 |
Mistral Nemotron
|
⚖️ | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 80 |
Mixtral 8x22b V0.1
|
⚖️ | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 81 |
Nemotron 3 Nano Omni 30b A3b Reasoning
|
⚖️ | NVIDIA Build | 30B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 82 |
Nemotron 3 Super 120b A12b
|
🚀 | NVIDIA Build | 120B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 83 |
Nemotron 3 Ultra 550b A55b
|
🚀 | NVIDIA Build | 550B | 1.0M | Free serverless inference for development; limits and model availability may change. | Open |
| 84 |
Nemotron 3.5 Lightning 30b A3b
|
⚖️ | NVIDIA Build | 30B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 85 |
Nemotron 4 340b Instruct
|
🚀 | NVIDIA Build | 340B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 86 |
Nemotron Nano 3 30b A3b
|
⚖️ | NVIDIA Build | 30B | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 87 |
Phi 3 Vision 128k Instruct
|
⚖️ | NVIDIA Build | ? | 131K | Free serverless inference for development; limits and model availability may change. | Open |
| 88 |
Phi 3.5 Moe Instruct
|
⚖️ | NVIDIA Build | ? | ? | Free serverless inference for development; limits and model availability may change. | Open |
| 89 |
Nemotron 3 Nano Omni 30b A3b Reasoning:Free
|
⚖️ | TokenRouter | 30B | ? | Free models and promotional access may be time-limited; availability is checked through the Models API. | Open |
| 90 |
Deepseek R1 Distill QWEN 32b
|
⚖️ | UnoRouter | 32B | 80K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 91 |
Gemini 1.5 Flash
|
⚖️ | UnoRouter | ? | 25K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 92 |
Gemini 3.5 Flash Lite
|
⚖️ | UnoRouter | ? | 1.0M | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 93 |
Gemini 3.6 Flash
|
⚖️ | UnoRouter | ? | 1.0M | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 94 |
Gemini Robotics Er 2 Preview
|
⚖️ | UnoRouter | ? | 131K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 95 |
Mistral Nemotron
|
⚖️ | UnoRouter | ? | 128K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 96 |
Mistral Small
|
⚖️ | UnoRouter | ? | 128K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 97 |
Nemotron 3 Nano Omni 30b A3b Reasoning
|
⚖️ | UnoRouter | 30B | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 98 |
Nemotron 3 Super 120b A12b
|
🚀 | UnoRouter | 120B | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 99 |
Nemotron 3 Ultra 550b A55b
|
🚀 | UnoRouter | 550B | 1.0M | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 100 |
Nemotron 3.5 Lightning
|
⚖️ | UnoRouter | ? | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 101 |
Nemotron 3.5 Lightning 30b A3b
|
⚖️ | UnoRouter | 30B | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 102 |
Nemotron Nano 12b V2 Vl
|
⚖️ | UnoRouter | 12B | 131K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 103 |
QWEN Sea Lion V4 32b It
|
⚖️ | UnoRouter | 32B | 33K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 104 |
QWEN Sea Lion V4.5 27b It
|
⚖️ | UnoRouter | 27B | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 105 | Qwen2.5 Coder 32b | ⚖️ | UnoRouter | 32B | 131K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 106 | Qwen3 | ⚖️ | UnoRouter | ? | 25K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 107 | Qwen3.5 397b A17b | 🚀 | UnoRouter | 397B | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 108 | Qwen3.6 35b A3b | ⚖️ | UnoRouter | 35B | 262K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 109 | Qwen3.8 27b | ⚖️ | UnoRouter | 27B | 66K | No-card free models use shared capacity and per-model limits; availability can change. | Open |
| 110 | Typhoon V2.5 30b A3b Instruct | ⚖️ | UnoRouter | 30B | 131K | No-card free models use shared capacity and per-model limits; availability can change. | Open |