# Cheapest Open-Weights Model APIs

Models whose weights are published. Prices as of 2026-11-22, USD per 1M tokens.

Models offered by at least 3 providers, ranked by the cheapest offer's 3:1 blended price ((3 × input + output) ÷ 4). Top 20 of 215.

| # | Model | Cheapest input | Cheapest output | Blended 3:1 | Providers |
| --- | --- | --- | --- | --- | --- |
| 1 | [GLM-5.3-Flash](https://sovyron.com/models/glm-5-3-flash/) | $0.0062 (JoinGonka) | $0.0185 (JoinGonka) | $0.0093 | 113 |
| 2 | [DeepSeek V4 Flash 0731](https://sovyron.com/models/deepseek-v4-flash-0731/) | $0.0062 (JoinGonka) | $0.0185 (JoinGonka) | $0.0093 | 53 |
| 3 | [MiniMax-M2.7](https://sovyron.com/models/minimax-m2-7/) | $0.0062 (JoinGonka) | $0.0185 (JoinGonka) | $0.0093 | 32 |
| 4 | [Llama 3.2 1B Instruct](https://sovyron.com/models/llama-3-2-1b-instruct/) | $0.01 (Inference) | $0.01 (Inference) | $0.01 | 3 |
| 5 | [Llama Prompt Guard 2 22M](https://sovyron.com/models/llama-prompt-guard-2-22m/) | $0.01 (Helicone) | $0.01 (Helicone) | $0.01 | 3 |
| 6 | [Llama 3.2 3B Instruct](https://sovyron.com/models/llama-3-2-3b-instruct/) | $0.02 (Eden AI) | $0.02 (Eden AI) | $0.02 | 6 |
| 7 | [Ling 3.0 Tiny](https://sovyron.com/models/ling-3-0-tiny/) | $0.015 (Kilo Gateway) | $0.045 (Kilo Gateway) | $0.0225 | 3 |
| 8 | [Llama-3.1-8B-Instruct](https://sovyron.com/models/llama-3-1-8b-instruct/) | $0.02 (Kilo Gateway) | $0.04 (Kilo Gateway) | $0.025 | 10 |
| 9 | [DeepSeek V4.1 Flash 1017](https://sovyron.com/models/deepseek-v4-1-flash-1017/) | $0.01 (Refuseless) | $0.08 (Refuseless) | $0.0275 | 73 |
| 10 | [Mistral Nemo](https://sovyron.com/models/mistral-nemo/) | $0.029 (OpenRouter) | $0.03 (OpenRouter) | $0.0293 | 8 |
| 11 | [GPT OSS 20B](https://sovyron.com/models/gpt-oss-20b/) | $0.015 (SOKKAN Router) | $0.075 (SOKKAN Router) | $0.03 | 38 |
| 12 | [Ling 3.0 Flash](https://sovyron.com/models/ling-3-0-flash/) | $0.021 (OpenRouter) | $0.063 (OpenRouter) | $0.0315 | 8 |
| 13 | [MiniMax-M3.1-Flash](https://sovyron.com/models/minimax-m3-1-flash/) | $0.02 (Refuseless) | $0.08 (Refuseless) | $0.035 | 34 |
| 14 | [Nemotron 3.5 Lightning 30B A3B](https://sovyron.com/models/nemotron-3-5-lightning-30b-a3b/) | $0.02 (Kilo Gateway) | $0.096 (Kilo Gateway) | $0.039 | 7 |
| 15 | [Ministral 3B](https://sovyron.com/models/ministral-3b/) | $0.04 (Azure) | $0.04 (Azure) | $0.04 | 7 |
| 16 | [Gemma 3 4B IT](https://sovyron.com/models/gemma-3-4b-it/) | $0.04 (Kilo Gateway) | $0.05 (Kilo Gateway) | $0.0425 | 6 |
| 17 | [Qwen3.5 4B](https://sovyron.com/models/qwen3-5-4b/) | $0.04 (EmpirioLabs AI) | $0.07 (EmpirioLabs AI) | $0.0475 | 3 |
| 18 | [GLM Flash Latest](https://sovyron.com/models/glm-flash-latest/) | $0.0399 (OpenRouter) | $0.0784 (OpenRouter) | $0.0495 | 3 |
| 19 | [Gemma 4 26B A4B](https://sovyron.com/models/gemma-4-26b-a4b/) | $0.0354 (NanoGPT) | $0.0943 (NanoGPT) | $0.0501 | 4 |
| 20 | [DeepSeek V4.1 Flash](https://sovyron.com/models/deepseek-v4-1-flash/) | $0.03 (engy) | $0.12 (engy) | $0.0525 | 82 |

HTML page: https://sovyron.com/cheapest/open-weights-models/
Methodology: https://sovyron.com/methodology/#blend
