Pythia Base: pricing, context window and benchmarks
Pythia Base is a open-weight model from EleutherAI. It offers a 16,384-token context window, costs $0.99 per million input tokens and $4.42 per million output tokens, and averages 55.9 across the index's benchmark suite.
Pricing
Pythia Base costs $0.99 per million input tokens and $4.42 per million output tokens. A blended 3:1 input-to-output mix works out to about $1.85 per million tokens. Cached input is listed at $0.11 per million tokens.
| Item | Price | Notes |
|---|---|---|
| Input | $0.99 | per 1M tokens |
| Output | $4.42 | per 1M tokens |
| Cached input | $0.11 | per 1M tokens |
| Blended 3:1 | $1.85 | estimate per 1M tokens |
| 1M in + 1M out | $5.41 | estimate |
Benchmarks
Scores below are modelled estimates on a 0–100 scale. Use them to compare relative position, not as official vendor results.
| Benchmark | Score | Rank | Percentile | Relative |
|---|---|---|---|---|
| MMLU-Pro | 60.5 | 524 | 13th | |
| GPQA Diamond | 51.0 | 510 | 15th | |
| SWE-bench Verified | 43.3 | 488 | 19th | |
| HumanEval | 59.6 | 509 | 15th | |
| MATH-500 | 55.7 | 512 | 15th | |
| AIME 2025 | 43.9 | 510 | 15th | |
| LiveCodeBench | 45.7 | 492 | 18th | |
| IFEval | 68.7 | 482 | 20th | |
| TruthfulQA | 62.1 | 503 | 16th | |
| HellaSwag | 71.9 | 503 | 16th | |
| MMMU | 52.9 | 521 | 13th | |
| Arena Elo | 1240.6 | 512 | 15th |
What Pythia Base is good for
Pythia Base is an open-weight model, so it can be self-hosted or run through a third-party inference provider. That matters for data residency, fine-tuning and long-run cost control. At $0.99 per million input tokens (hosted list price) and a 16,384-token context window, it is a practical option for teams that want more control than a closed API allows.
Strengths
- Input price of $0.99 per million tokens is at or below the index median.
- Open weights make self-hosting, fine-tuning and data-residency review possible.
Trade-offs
- Context window of 16,384 tokens is small for long-document work.
- Average benchmark score is below the index median; test it on your own task.
How it compares
| Model | Provider | Context | Input / 1M | Avg score |
|---|---|---|---|---|
| Llama 3 Hyperbolic 8B | Hyperbolic | 16384 | $0.92 | 56.0 |
| Llama 3 Replicate 7B | Replicate | 32768 | $0.91 | 56.0 |
| Gemma 3 7B | Google DeepMind | 32768 | $0.90 | 55.9 |
| Pythia Base | EleutherAI | 16384 | $0.99 | 55.9 |
| Yi-Lightning 7B | 01.AI | 32768 | $1.00 | 55.9 |
| Embed 4 Mini | Cohere | 4096 | $2.48 | 55.8 |
FAQ
How much does Pythia Base cost?
Pythia Base is listed at $0.99 per million input tokens and $4.42 per million output tokens. Cached input is $0.11 per million tokens. A blended 3:1 input-to-output mix is about $1.85 per million tokens.
What is the context window of Pythia Base?
Pythia Base has a context window of 16,384 tokens and can emit up to 16,384 output tokens per request in this index.
Is Pythia Base open weights?
Yes. Pythia Base is listed as an open-weight model under a open weights licence.
How does Pythia Base score on benchmarks?
Pythia Base averages 55.9 across the scored benchmarks in this index, ranking #508 of 600 models. Its strongest relative result is Arena Elo at 1240.6.
What is Pythia Base best for?
Pythia Base is an open-weight model, so it can be self-hosted or run through a third-party inference provider. That matters for data residency, fine-tuning and long-run cost control. At $0.99 per million input tokens (hosted list price) and a 16,384-token context window, it is a practical option for teams that want more control than a closed API allows.