README.md

---
license: mit
language:
  - en
tags:
  - llm
  - pricing
  - cost
  - openai
  - anthropic
  - google
  - meta
  - reference
size_categories:
  - n<1K
pretty_name: LLM Pricing Table
configs:
  - config_name: default
    data_files:
      - split: train
        path: data.jsonl
---

# LLM Pricing Table

Per-1k-token input/output costs for the major LLM models, in a single loadable JSONL. Useful for cost-estimator dashboards, budget enforcement, and ROI analysis.

```python
from datasets import load_dataset
ds = load_dataset("mukunda1729/model-pricing-table", split="train")
prices = {row["model"]: row for row in ds}
print(prices["claude-sonnet-4-6"]["input_per_1k_tokens_usd"])  # 0.003
```

## Schema

| Field | Type | Notes |
|---|---|---|
| `model` | `str` | Canonical model identifier |
| `provider` | `str` | `openai` / `anthropic` / `google` / `meta` / `mistral` / `deepseek` |
| `input_per_1k_tokens_usd` | `float` | USD per 1,000 input tokens |
| `output_per_1k_tokens_usd` | `float` | USD per 1,000 output tokens |
| `context_window` | `int` | Max tokens (input + output) |
| `modality` | `str` | `text` / `multimodal` |

## Data freshness

Snapshot as of 2026-04-27. Provider prices change — always cross-reference the official pricing page before relying on these in production billing.

Sister tooling: [`llm-cost-guard-py`](https://pypi.org/project/llm-cost-guard-py/) and [`@mukundakatta/llm-cost-guard`](https://www.npmjs.com/package/@mukundakatta/llm-cost-guard) consume this table directly.

Part of [The Agent Reliability Stack](https://mukundakatta.github.io/agent-stack/).

## License

MIT.