LLM list-price dataset
33
models priced
2026-07-15
verified by a human
26d
since verification · gate 90d
These are list rates, not invoices. Useful for planning and comparison. A real bill differs with prompt caching, batch tiers and negotiated pricing — which is why TryTokka reconciles estimates against the provider’s own invoice instead of trusting a rate table.
All figures in USD per 1,000,000 tokens.
OpenAI
| Model | Input | Output | Cached input |
|---|---|---|---|
| gpt-3.5-turbo | $0.50 | $1.50 | — |
| gpt-4 | $30.00 | $60.00 | — |
| gpt-4-turbo | $10.00 | $30.00 | — |
| gpt-4-turbo-preview | $10.00 | $30.00 | — |
| gpt-4.1 | $2.00 | $8.00 | $0.500 |
| gpt-4.1-mini | $0.40 | $1.60 | $0.100 |
| gpt-4.1-nano | $0.10 | $0.40 | $0.025 |
| gpt-4o | $2.50 | $10.00 | $1.250 |
| gpt-4o-2024-08-06 | $2.50 | $10.00 | $1.250 |
| gpt-4o-2024-11-20 | $2.50 | $10.00 | $1.250 |
| gpt-4o-mini | $0.15 | $0.60 | $0.075 |
| gpt-4o-mini-2024-07-18 | $0.15 | $0.60 | $0.075 |
| gpt-5 | $1.25 | $10.00 | $0.125 |
| gpt-5-mini | $0.25 | $2.00 | $0.125 |
| gpt-5-nano | $0.05 | $0.40 | $0.025 |
| o1 | $15.00 | $60.00 | $7.500 |
| o1-mini | $1.10 | $4.40 | $0.550 |
| o1-preview | $15.00 | $60.00 | $7.500 |
| o3 | $2.00 | $8.00 | $0.500 |
| o3-mini | $1.10 | $4.40 | $0.550 |
| o4-mini | $1.10 | $4.40 | $0.550 |
| text-embedding-3-large | $0.13 | — | — |
| text-embedding-3-small | $0.02 | — | — |
| text-embedding-ada-002 | $0.10 | — | — |
Anthropic
| Model | Input | Output | Cached input |
|---|---|---|---|
| claude-3-5-haiku | $0.80 | $4.00 | $0.080 |
| claude-3-5-opus | $15.00 | $75.00 | $1.500 |
| claude-3-5-sonnet | $3.00 | $15.00 | $0.300 |
| claude-3-7-sonnet | $3.00 | $15.00 | $0.300 |
| claude-3-haiku | $0.25 | $1.25 | $0.030 |
| claude-3-opus | $15.00 | $75.00 | $1.500 |
| claude-3-sonnet | $3.00 | $15.00 | $0.300 |
| claude-opus-4 | $15.00 | $75.00 | $1.500 |
| claude-sonnet-4 | $3.00 | $15.00 | $0.300 |
Methodology
- Rates are read by hand from each provider’s published pricing page and pinned in source control.
- CI fails the build if a pinned rate drifts from OpenRouter, or if the verification date is older than 90 days.
- Cached input is the prompt-cache read rate where the provider offers one, and “—” where it does not.
- Output shows “—” for embedding models, which produce no billable output tokens. It is never shown as $0.00, which would read as free output.
- Nothing is auto-updated from a provider page. A human always reads the change, because a confidently wrong rate is worse than an old one.
Licence & citation
Published under CC BY 4.0. Suggested citation:
TryTokka, LLM List-Price Dataset, Acsaven. https://trytokka.com/llm-rate-dataset
Questions
- Can I reuse this dataset?
- Yes — CC BY 4.0, with attribution to TryTokka. The prices are published by the providers; what is offered here is the pinned, dated and drift-checked compilation.
- Are these what I will actually be billed?
- No. These are published list rates, useful for planning and comparison. A real invoice differs with prompt caching, batch tiers and negotiated pricing — which is exactly why TryTokka reconciles estimates against the provider’s own invoice rather than trusting a rate table.
- How do you keep it current?
- Rates are pinned in source control with the date a human verified them (2026-07-15). CI fails the build if a pin drifts from OpenRouter, or if the verification is older than 90 days. Nothing is auto-updated from a provider page — a confidently wrong rate is worse than an old one.