OpenAI API Pricing 2026
The most widely-used LLM API. GPT-5.5 and GPT-5.4 are the current frontier tiers; o3-pro and o4-mini lead the reasoning tier. OSS 120B offers open-source quality at near-free pricing.
Pricing verified 2026-06-06. Sourced from platform.openai.com.
Get OpenAI API access →OpenAI Model Pricing
Prices in USD per 1M tokens
| Model | Input / 1M | Output / 1M | Context |
|---|---|---|---|
GPT-5.5 Latest OpenAI frontier model; 272K context standard tier | $5 | $30 | 272,000 |
GPT-5.5 Pro GPT-5.5 Pro; 1M context; max capability tier | $30 | $180 | 1,050,000 |
GPT-5.4 Strong coding & professional workloads; 272K context standard tier | $2.5 | $15 | 272,000 |
GPT-5.4 Mini Best mini model for coding, computer use, and subagents | $0.75 | $4.5 | 272,000 |
GPT-5.4 Nano Ultra-low-cost GPT-5.4 tier for classification and routing | $0.2 | $1.25 | 272,000 |
GPT-5.4 Pro GPT-5.4 Pro; 1M context; enterprise workloads | $30 | $180 | 1,050,000 |
GPT-5 Pro GPT-5 Pro; 400K context; heavy inference | $15 | $120 | 400,000 |
o3 Deep Research o3 with deep-research mode; extended reasoning budget | $10 | $40 | 200,000 |
o4 Mini Deep Research o4-mini deep-research; cost-effective research tasks | $2 | $8 | 200,000 |
GPT-5 Codex GPT-5 Codex; 400K context; coding-optimized | $1.25 | $10 | 400,000 |
GPT-5 Mini Previous-gen budget model; GPT-5.4 Mini is the current mini tier | $0.25 | $2 | 128,000 |
GPT-5 Previous flagship; use GPT-5.5 for new projects | $1.25 | $10 | 128,000 |
GPT-5 Nano GPT-5 Nano; 400K context; ultra-cheap classification & routing | $0.05 | $0.4 | 400,000 |
gpt-oss-120b OpenAI open-source 120B; runs on Groq/Together at low cost | $0.04 | $0.18 | 131,072 |
o3 Pro o3 Pro; extended thinking compute; top-tier reasoning | $20 | $80 | 200,000 |
GPT-4.1 Previous production tier with 1M context; superseded by GPT-5.4 | $2 | $8 | 1,000,000 |
GPT-4.1 Mini Previous mid-tier; 1M context window | $0.4 | $1.6 | 1,000,000 |
GPT-4.1 Nano Previous budget tier; GPT-5.4 Nano is cheaper for new workloads | $0.1 | $0.4 | 1,000,000 |
o3 Flagship reasoning; use when o4-mini accuracy is insufficient | $2 | $8 | 200,000 |
o4-mini Best-value reasoning model; chain-of-thought built in | $1.1 | $4.4 | 200,000 |
o1-pro Auto-added 2026-06 | $150 | $600 | 200,000 |
o3 Mini Auto-added 2026-06 | $1.1 | $4.4 | 200,000 |
o1 Auto-added 2026-06 | $15 | $60 | 200,000 |
GPT-4o-mini Auto-added 2026-06 | $0.15 | $0.6 | 128,000 |
Estimated Monthly Cost (70% input / 30% output split)
| Model | 1M tokens/mo | 10M tokens/mo | 100M tokens/mo | 1B tokens/mo |
|---|---|---|---|---|
| GPT-5.5 | $12.50 | $125 | $1,250 | $12,500 |
| GPT-5.5 Pro | $75.00 | $750 | $7,500 | $75,000 |
| GPT-5.4 | $6.25 | $62.50 | $625 | $6,250 |
| GPT-5.4 Mini | $1.87 | $18.75 | $188 | $1,875 |
| GPT-5.4 Nano | $0.515 | $5.15 | $51.50 | $515 |
| GPT-5.4 Pro | $75.00 | $750 | $7,500 | $75,000 |
| GPT-5 Pro | $46.50 | $465 | $4,650 | $46,500 |
| o3 Deep Research | $19.00 | $190 | $1,900 | $19,000 |
| o4 Mini Deep Research | $3.80 | $38.00 | $380 | $3,800 |
| GPT-5 Codex | $3.88 | $38.75 | $388 | $3,875 |
| GPT-5 Mini | $0.775 | $7.75 | $77.50 | $775 |
| GPT-5 | $3.88 | $38.75 | $388 | $3,875 |
| GPT-5 Nano | $0.155 | $1.55 | $15.50 | $155 |
| gpt-oss-120b | $0.082 | $0.820 | $8.20 | $82.00 |
| o3 Pro | $38.00 | $380 | $3,800 | $38,000 |
| GPT-4.1 | $3.80 | $38.00 | $380 | $3,800 |
| GPT-4.1 Mini | $0.760 | $7.60 | $76.00 | $760 |
| GPT-4.1 Nano | $0.190 | $1.90 | $19.00 | $190 |
| o3 | $3.80 | $38.00 | $380 | $3,800 |
| o4-mini | $2.09 | $20.90 | $209 | $2,090 |
| o1-pro | $285 | $2,850 | $28,500 | $285,000 |
| o3 Mini | $2.09 | $20.90 | $209 | $2,090 |
| o1 | $28.50 | $285 | $2,850 | $28,500 |
| GPT-4o-mini | $0.285 | $2.85 | $28.50 | $285 |
Frequently Asked Questions
How much does OpenAI LLM API cost?
OpenAI offers 24 models ranging from $0.040/1M to $150.00/1M input tokens. The most widely-used LLM API. GPT-5.5 and GPT-5.4 are the current frontier tiers; o3-pro and o4-mini lead the reasoning tier. OSS 120B offers open-source quality at near-free pricing.
Is OpenAI cheaper than self-hosting?
For low-volume workloads (under 100M tokens/month), cloud APIs like OpenAI are almost always cheaper than purchasing and maintaining GPU hardware. Use our calculator to find the exact break-even point for your usage.