Cheapest LLM APIs in 2026: full ranking
All 600 models are ranked by blended 3:1 price, lowest first, using input and output prices from the index.
What the data shows
The spread runs from about $0.073 to $43.66 per million tokens on a blended 3:1 basis. The median across the set is $7.65.
The top of the list is dominated by small and open-weight models. That is expected: price and capability trade off, and the cheapest model is rarely the strongest on benchmarks.
How to read the ranking
Blended price weights input and output at 3:1. If your workload is output-heavy, the ranking can change, so check the output price column as well.
- Filter by context window before comparing price.
- Check whether the model is open weights or API-only.
- Test the cheapest candidate on your own task before switching.
- Keep a fallback model for quality-sensitive requests.
Blended 3:1 price, lowest first
| # | Model | Provider | Input / 1M | Output / 1M | Context | Avg score |
|---|---|---|---|---|---|---|
| 1 | SmolLM3 1.5B | Hugging Face | $0.040 | $0.17 | 8,192 | 36.8 |
| 2 | GPT-NeoX 1.5B | EleutherAI | $0.060 | $0.32 | 32,768 | 36.4 |
| 3 | Pythia 7B | EleutherAI | $0.10 | $0.36 | 8,192 | 40.7 |
| 4 | Nemotron Nano 1.5B | NVIDIA | $0.10 | $0.48 | 8,192 | 41.5 |
| 5 | Zephyr 1.5B | Hugging Face | $0.10 | $0.51 | 32,768 | 38.0 |
| 6 | Pythia 1.5B | EleutherAI | $0.10 | $0.54 | 4,096 | 35.8 |
| 7 | StableLM 2 1.5B | Stability AI | $0.13 | $0.46 | 16,384 | 38.3 |
| 8 | Arctic 1.5B | Snowflake | $0.15 | $0.82 | 32,768 | 43.2 |
| 9 | Pythia 8B | EleutherAI | $0.16 | $0.79 | 32,768 | 41.4 |
| 10 | SmolLM3 8B | Hugging Face | $0.21 | $0.76 | 32,768 | 44.0 |
| 11 | GPT-NeoX 7B | EleutherAI | $0.20 | $0.89 | 8,192 | 42.4 |
| 12 | Zephyr 8B | Hugging Face | $0.22 | $0.84 | 8,192 | 43.0 |
| 13 | Molmo 1.5B | Allen Institute for AI | $0.21 | $0.93 | 8,192 | 42.8 |
| 14 | SmolLM3 7B | Hugging Face | $0.19 | $1.03 | 16,384 | 44.3 |
| 15 | Zephyr 7B | Hugging Face | $0.23 | $0.93 | 32,768 | 42.5 |
| 16 | GPT-NeoX 8B | EleutherAI | $0.25 | $1.04 | 16,384 | 43.7 |
| 17 | Phi-3.5 1.5B | Microsoft | $0.25 | $1.05 | 32,768 | 44.5 |
| 18 | Mixtral Turbo 1.5B | Together AI | $0.31 | $1.19 | 4,096 | 47.6 |
| 19 | DBRX 1.5B | Databricks | $0.32 | $1.22 | 16,384 | 47.0 |
| 20 | OLMo 2 1.5B | Allen Institute for AI | $0.28 | $1.41 | 16,384 | 44.0 |
| 21 | Mixtral Groq 1.5B | Groq | $0.34 | $1.24 | 4,096 | 46.1 |
| 22 | StableLM 2 7B | Stability AI | $0.28 | $1.47 | 8,192 | 44.3 |
| 23 | Nemotron Nano 7B | NVIDIA | $0.35 | $1.30 | 8,192 | 47.5 |
| 24 | Mixtral 1.5B | Mistral AI | $0.36 | $1.50 | 16,384 | 47.7 |
| 25 | Mistral Small 3 1.5B | Mistral AI | $0.42 | $1.58 | 32,768 | 49.2 |
| 26 | StableLM 2 8B | Stability AI | $0.38 | $1.78 | 32,768 | 47.0 |
| 27 | Nemotron Nano 8B | NVIDIA | $0.46 | $1.64 | 8,192 | 48.1 |
| 28 | Gemma 3 1.5B | Google DeepMind | $0.40 | $1.82 | 32,768 | 48.9 |
| 29 | Llama 3 Hyperbolic 1.5B | Hyperbolic | $0.45 | $1.70 | 32,768 | 48.4 |
| 30 | Llama 3 Modal 1.5B | Modal | $0.42 | $1.80 | 4,096 | 49.7 |
| 31 | Phi-4 1.5B | Microsoft | $0.42 | $1.90 | 8,192 | 48.1 |
| 32 | Llama 3.3 Groq 1.5B | Groq | $0.46 | $1.80 | 4,096 | 50.0 |
| 33 | Llama 3 Baseten 1.5B | Baseten | $0.42 | $2.01 | 4,096 | 48.2 |
| 34 | Llama 3 Lambda 1.5B | Lambda | $0.46 | $1.91 | 16,384 | 48.4 |
| 35 | Llama 3 Novita 1.5B | Novita AI | $0.41 | $2.09 | 4,096 | 49.1 |
| 36 | Llama 3 DeepInfra 1.5B | DeepInfra | $0.48 | $1.91 | 32,768 | 48.8 |
| 37 | Yi-Large 1.5B | 01.AI | $0.42 | $2.13 | 8,192 | 46.8 |
| 38 | Molmo 7B | Allen Institute for AI | $0.42 | $2.22 | 32,768 | 49.3 |
| 39 | Llama 3 Anyscale 1.5B | Anyscale | $0.43 | $2.27 | 16,384 | 48.9 |
| 40 | Codestral 1.5B | Mistral AI | $0.52 | $2.03 | 8,192 | 49.9 |
| 41 | Molmo 8B | Allen Institute for AI | $0.44 | $2.29 | 32,768 | 49.9 |
| 42 | Llama 3.1 Turbo 1.5B | Together AI | $0.46 | $2.24 | 16,384 | 49.8 |
| 43 | Seed-OSS 1.5B | ByteDance | $0.45 | $2.36 | 4,096 | 49.8 |
| 44 | Llama 3.1 1.5B | Meta AI | $0.53 | $2.14 | 4,096 | 50.7 |
| 45 | Llama 3 Replicate 1.5B | Replicate | $0.51 | $2.28 | 4,096 | 49.3 |
| 46 | Arctic 7B | Snowflake | $0.49 | $2.38 | 8,192 | 50.3 |
| 47 | Command R+ 1.5B | Cohere | $0.51 | $2.35 | 32,768 | 50.0 |
| 48 | Llama 3.1 Cerebras 1.5B | Cerebras | $0.48 | $2.46 | 4,096 | 50.2 |
| 49 | Nemotron 4 1.5B | NVIDIA | $0.49 | $2.67 | 32,768 | 50.8 |
| 50 | Yi-Lightning 1.5B | 01.AI | $0.54 | $2.52 | 4,096 | 50.0 |
Methodology
This story uses the AI & LLM Index dataset: 600 models, 40 providers and 12 benchmarks. Prices are listed per million tokens. Scores are modelled estimates on a 0–100 scale and are not official benchmark results. Figures are rounded for readability.
Related
FAQ
What does the Cheapest LLM APIs in 2026: full ranking data show?
All 600 models are ranked by blended 3:1 price, lowest first, using input and output prices from the index.
Are the numbers official?
No. Prices and scores are compiled for this demo index. Benchmark values are modelled estimates, not official results. Verify figures with the provider or benchmark owner before making decisions.
How should I use this ranking?
Use it to form a shortlist. Check the specific model pages, compare price and context, and test the shortlist on your own examples before committing.