Tools & price references
LLM price comparison.
Compare selected OpenAI, Claude and Gemini prices in the same unit: USD per million text tokens. Filter by provider and sort by price.
Updated
Price snapshot checked 2026-10-28. Sources are linked per model.
12 selected models
| Model | Input / 1M | Cache read / 1M | Output / 1M | Batch input / 1M | Source & scope |
|---|---|---|---|---|---|
| GPT-6.1 SolOpenAI | $2 | $0.1 | $10 | $1 | Official source2026-10-28 Higher tier above 272,000 input tokens. |
| GPT-6.1 TerraOpenAI | $1 | $0.05 | $5 | $0.5 | Official source2026-10-28 Higher tier above 272,000 input tokens. |
| GPT-6 LunaOpenAI | $0.1 | $0.01 | $0.5 | $0.05 | Official source2026-10-28 Higher tier above 272,000 input tokens. |
| GPT-6 AstraOpenAI | $10 | $1 | $50 | $5 | Official source2026-10-28 Higher tier above 272,000 input tokens. |
| Claude Sonnet 5.5Claude | $2 | $0.1 | $10 | $1 | Official source2026-10-28 Base prompt tier. |
| Claude Opus 5.5Claude | $4 | $0.2 | $20 | $2 | Official source2026-10-28 Base prompt tier. |
| Claude Haiku 5.5Claude | $0.1 | $0.01 | $0.5 | $0.05 | Official source2026-10-28 Higher tier above 100,000 input tokens. |
| Claude Fable 5.1Claude | $10 | $0.25 | $50 | $5 | Official source2026-10-28 Base prompt tier. |
| Gemini 4 ArgonGemini | $2 | $0.1 | $10 | $1 | Official source2026-10-28 Base prompt tier. |
| Gemini 3.8 FlashGemini | $0.75 | $0.075 | $3.75 | $0.375 | Official source2026-10-28 Base prompt tier. |
| Gemini 3.5 Flash-LiteGemini | $0.3 | $0.03 | $2.5 | $0.15 | Official source2026-10-28 Base prompt tier. |
| Gemini 3.1 Pro PreviewGemini | $2 | $0.2 | $12 | $1 | Official source2026-10-28 Higher tier above 200,000 input tokens. |
USD per million text tokens. Selected models, not a complete catalog. Base prompt tier shown. Cache writes/storage and other charges may apply. Scroll the table sideways on mobile.
Price is one input to model selection
This table compares billing rates, not model quality, latency or reliability. Test a representative sample and count accepted results, retries and review time before choosing a model.
Match the workload
A short prompt and a long answer have a different cost mix from a large document and a short extraction. Use your actual input/output ratio.
Calculate your workloadMatch the billing mode
Batch, cache writes, context tiers and tool fees can change the comparison. Do not compare a standard rate with a discounted rate as though they are equivalent.
Check billing assumptions