Pricing Comparison
Per-token pricing, side by side
Sort any numeric column to rank models by cost. Prices are USD per 1M tokens.
Provider
Search
Tier
| Provider | |||||
|---|---|---|---|---|---|
| OpenAI | GPT-6 Luna | 1.1M | $0.100 | $0.500 | $0.010 |
| Anthropic | Claude Haiku 5.5 | 1M | $0.100 | $0.500 | $0.010 |
| Mistral | Mistral Small 4 | 256K | $0.150 | $0.600 | $0.015 |
| DeepSeek | DeepSeek-V4.1-Flash | 1M | $0.150 | $0.600 | $0.003 |
| Gemini 3.5 Flash-Lite | 1M | $0.300 | $2.50 | $0.030 | |
| DeepSeek | DeepSeek-V4-Pro | 1M | $0.660 | $1.98 | $0.022 |
| Mistral | Mistral Large 4 | 1M | $0.680 | $2.09 | $0.070 |
| Gemini 3.8 Flash | 1M | $0.750 | $3.75 | $0.075 | |
| Mistral | Mistral Medium 3.5 | 256K | $1.50 | $7.50 | $0.150 |
| OpenAI | GPT-6.1 Sol | 1.1M | $2.00 | $10.00 | $0.100 |
| OpenAI | GPT-5.6 Terra | 1.1M | $2.00 | $12.00 | $0.200 |
| Anthropic | Claude Sonnet 5.5 | 1M | $2.00 | $10.00 | $0.100 |
| Gemini 4 Argon | 1.0M | $2.00 | $10.00 | $0.100 | |
| xAI | Grok 4.7 | 500K | $2.00 | $6.00 | $0.500 |
| Anthropic | Claude Opus 5.5 | 1M | $4.00 | $20.00 | $0.200 |
| OpenAI | GPT-6 Astra | 1.1M | $10.00 | $50.00 | $1.00 |
| Anthropic | Claude Fable 5.1 | 1M | $10.00 | $50.00 | $0.250 |
17 of 17 models shown · click any row for full detail · prices in USD per 1M tokens
Methodology
- Prices are pulled directly from each provider's official pricing page as of the "data as of" date in the footer — click any row for the source link.
- Figures are USD per 1,000,000 tokens at published list price; negotiated, enterprise, or volume-discount rates aren't reflected.
- Tier (frontier / mid-range / lightweight) is assigned by this dashboard based on each model's price and positioning within its own provider's lineup, not a third-party standard.
- Cached-input price reflects the provider's own published prompt-caching rate where one exists; where a provider doesn't publish one directly, it's derived from Artificial Analysis's published cache-hit discount for that model instead (flagged in the model's detail drawer). It's blank where neither source shows a distinct cached rate.
- LLM pricing changes frequently — treat this as a snapshot and verify against the source before relying on it.