LLM Pricing Comparison (January 2026)
Prices change frequently. Here's the current pricing from official API documentation, sorted by input cost:
| Model | Input (per 1M tokens) | Output (per 1M tokens) |
|---|---|---|
| GPT-4o mini OpenAI | $0.15 | $0.60 |
| DeepSeek V3 DeepSeek | $0.27 | $1.10 |
| Gemini 2.5 Flash Google | $0.30 | $2.50 |
| DeepSeek R1 DeepSeek (reasoning) | $0.55 | $2.19 |
| Claude Haiku 4.5 Anthropic | $1.00 | $5.00 |
| Gemini 2.5 Pro Google | $1.25 | $10.00 |
| GPT-4o OpenAI | $2.50 | $10.00 |
| o1-mini OpenAI (reasoning) | $3.00 | $12.00 |
| Claude Sonnet 4.5 Anthropic | $3.00 | $15.00 |
| Claude Opus 4.5 Anthropic | $5.00 | $25.00 |
| o1 OpenAI (reasoning) | $15.00 | $60.00 |
Key insight: DeepSeek V3 is roughly 10x cheaper than GPT-4o for input tokens. For cost-sensitive workloads where quality is acceptable, this is a significant optimization lever.
Cost Calculator
Calculate Your Monthly LLM Costs
5 Ways to Reduce LLM Costs
1. Use smaller models for simple tasks 50-90% savings
GPT-4o mini handles classification, extraction, and simple Q&A just as well as GPT-4o at 94% less cost. Route complex reasoning to expensive models, everything else to cheap ones.
2. Enable prompt caching Up to 90% savings
Anthropic charges only 10% for cached prompt tokens. If you have repeated system prompts or context, caching can dramatically reduce input costs. OpenAI and Google offer similar features.
3. Batch non-urgent requests 50% savings
OpenAI's Batch API offers 50% off for requests that can wait up to 24 hours. Perfect for content generation, data processing, or overnight analysis jobs.
4. Consider DeepSeek for cost-sensitive workloads 80-90% savings
DeepSeek V3 benchmarks competitively with GPT-4o on many tasks at 1/10th the input cost. Test it on your use case—you might be surprised.
5. Monitor and set budget alerts Prevent overruns
Most teams discover cost issues after receiving a surprise invoice. Use tools like Helicone or your provider's built-in usage dashboards to set alerts before you hit limits. See our Best LLM Monitoring Tools guide for platform comparisons.
Frequently Asked Questions
How much does GPT-4 cost per token?
GPT-4o costs $2.50 per million input tokens and $10.00 per million output tokens. That's $0.0000025 per input token and $0.00001 per output token. GPT-4o mini is much cheaper at $0.15/$0.60 per million tokens.
What is the cheapest LLM API in 2026?
DeepSeek V3 is currently the cheapest major LLM API at $0.27 per million input tokens. Among major US providers, GPT-4o mini ($0.15 input) is the cheapest option from OpenAI, while Gemini 2.5 Flash ($0.30 input) is competitive from Google.
How can I reduce my OpenAI API costs?
The most effective methods are: use GPT-4o mini instead of GPT-4o for simpler tasks (94% cheaper), enable prompt caching for repeated context, use the Batch API for non-urgent requests (50% off), and optimize your prompts to use fewer tokens.
Is Claude cheaper than GPT-4?
Claude Sonnet 4.5 ($3.00/$15.00) is slightly more expensive than GPT-4o ($2.50/$10.00) for most workloads. However, Claude Haiku ($1.00/$5.00) is competitive with GPT-4o mini for lightweight tasks. The best choice depends on your quality requirements and use case.
What is prompt caching?
Prompt caching stores frequently-used prompt prefixes so you don't pay full price for repeated context. Anthropic charges only 10% for cached tokens—so if you have a 2,000 token system prompt that's repeated on every request, caching saves 90% on those tokens. Enable it by structuring prompts with static content first.
Need help optimizing your LLM costs?
Integrity Studio helps teams monitor AI spending and implement cost optimization strategies.
Start Monitoring Your LLM CostsLast verified: January 30, 2026. Pricing from OpenAI, Anthropic, Google AI, and DeepSeek.