EleutherAI API: models, pricing and limits
A practical guide to eleutherai, using data from the AI & LLM Index. The goal is a shortlist and a decision, not a tour of every option.
The EleutherAI lineup
EleutherAI has 14 models in the index. Open research collective behind Pythia and GPT-NeoX. The lineup spans 2 categories, from open-weight to small open-weight.
The strongest model by average score is gpt-neox-instruct. The cheapest by input price is GPT-NeoX 1.5B.
Pricing
Across the lineup, the average input price is $0.83 per million tokens and the average output price is $3.56. That average hides a wide range: individual models run from $0.060 to $1.57 on input.
Use the average as a starting point, then check the specific model you plan to deploy. Output price and cached input pricing can change the ranking for long or repetitive workloads.
Strengths and limits
14 of the 14 models are open weights. That gives you a self-hosting path for part of the lineup, while the proprietary models remain API-only.
The best benchmark result in the lineup is 62.6 on the index average. Compare that with the best model from other providers before treating EleutherAI as the default.
How to choose
- Start with the task, not the brand.
- Pick the cheapest model that clears your evaluation set.
- Keep a second provider as a fallback for outages and price changes.
- Check licence and data-handling terms for every model you shortlist.
EleutherAI models at a glance
| Model | Input / 1M | Context | Avg score |
|---|---|---|---|
| GPT-NeoX Instruct | $1.52 | 16,384 | 62.6 |
| GPT-NeoX | $1.57 | 16,384 | 61.7 |
| Pythia Instruct | $1.41 | 8,192 | 60.8 |
| GPT-NeoX Chat | $1.49 | 16,384 | 60.1 |
| Pythia | $1.29 | 32,768 | 59.8 |
| GPT-NeoX Base | $1.22 | 16,384 | 59.7 |
| Pythia Chat | $1.31 | 4,096 | 59.1 |
| Pythia Base | $0.99 | 16,384 | 55.9 |
| GPT-NeoX 8B | $0.25 | 16,384 | 43.7 |
| GPT-NeoX 7B | $0.20 | 8,192 | 42.4 |
Related
FAQ
Which EleutherAI model should I start with?
Start with gpt-neox-instruct for quality-sensitive work and GPT-NeoX 1.5B for high-volume tasks.
What is the most important factor in eleutherai?
Fit to the task beats raw benchmark rank. Define the failure modes, pick the benchmarks that resemble them, then test a small shortlist on your own examples.
How many models should I evaluate?
Three is usually enough: a quality candidate, a value candidate and a fallback. More candidates add work without changing the decision.
How do I control cost?
Compare blended 3:1 price, watch output tokens, use caching where available and route simple requests to a cheaper model.