GPT-5.6 API Pricing: Sol, Terra, Luna Costs Compared
OpenAI's latest model family has three tiers. Here's exactly what each costs, how they compare to competitors, and which one you should use.
Pricing update (August 24, 2026): OpenAI is temporarily charging qualifying GPT-5.6 Sol API usage at $4/M input, $0.40/M cached input, and $20/M output. The published list price remains $5/$0.50/$30. OpenAI says the promotional rate is available through at least November 21, 2026; included-plan usage and legacy credit metering can differ. The historical comparisons below retain list pricing so the temporary promotion is not mistaken for a permanent cut.
Availability update (August 25, 2026): Sol, Terra, and Luna are available inside Kiro as coding-agent models. Separately, AWS has made Terra and Luna generally available through Amazon Bedrock in AWS GovCloud (US-East and US-West). That GovCloud path uses Bedrock access and billing; it is not the direct OpenAI API. AWS did not include Sol in that specific GovCloud announcement.
OpenAI released GPT-5.6 in July 2026 with a new three-tier structure: Sol (premium), Terra (balanced), and Luna (budget). All three share a massive 1.05M token context window — the largest available via any API.
Here's the quick version:
GPT-5.6 Pricing Breakdown
| Model | Input/1M | Output/1M | Context | Tier |
|---|---|---|---|---|
| GPT-5.6 Sol | $4.00 promo $5.00 list; $0.40 cached promo |
$20.00 promo $30.00 list |
1.05M | Premium |
| GPT-5.6 Terra | $2.50 | $15.00 | 1.05M | Mid |
| GPT-5.6 Luna | $1.00 | $6.00 | 1.05M | Budget |
Key detail: Output tokens cost 6x more than input tokens across all GPT-5.6 models. This is the same ratio as GPT-5.5, but the absolute costs are lower for Terra and Luna.
GPT-5.6 vs Every Competitor
| Model | Input/1M | Output/1M | vs GPT-5.6 Sol | Context |
|---|---|---|---|---|
| GPT-5.6 Sol | $5.00 | $30.00 | — | 1.05M |
| GPT-5.6 Terra | $2.50 | $15.00 | 60% cheaper input, 60% cheaper output | 1.05M |
| GPT-5.6 Luna | $1.00 | $6.00 | 96% cheaper input, 96% cheaper output | 1.05M |
| Claude Opus 5 | $5.00 | $25.00 | Same input, 17% cheaper output | 1M |
| Claude Sonnet 5 | $2.00 | $10.00 | 60% cheaper input, 67% cheaper output | 1M |
| Gemini 3.1 Pro | $2.00 | $12.00 | 60% cheaper input, 60% cheaper output | 1M |
| GPT-5 | $1.25 | $10.00 | 75% cheaper input, 67% cheaper output | 272K |
| DeepSeek V4 Pro | $0.66 | $1.98 | 91% cheaper input, 97% cheaper output | 1M |
| Mistral Large 3 | $0.50 | $1.50 | 90% cheaper input, 95% cheaper output | 262K |
Key insight: GPT-5.6 Sol is now cheaper than GPT-5.5 ($4/$20 vs $5/$30) but with a 5% larger context window (1.05M vs 1M). If you're already on GPT-5.5, GPT-5.6 Sol is a free upgrade with more context. GPT-5.6 Terra is the sweet spot — 60% cheaper than Sol with the same context window.
Real-World Cost Scenarios
Scenario 1: Customer Support Chatbot (1,000 requests/day)
Average: 1,500 input tokens, 400 output tokens per request. 30 days/month.
Monthly Chatbot Cost
Verdict: GPT-5.6 Luna ($59.40/mo) is the cheapest OpenAI option for chatbots. But DeepSeek V4 Flash ($9.66/mo) is 84% cheaper if you don't need OpenAI's ecosystem. For most chatbots, Claude Sonnet 5 ($108/mo) offers the best quality-to-cost ratio.
Scenario 2: Code Generation (200 requests/day)
Average: 3,000 input tokens, 1,200 output tokens per request. 30 days/month.
Monthly Code Generation Cost
Verdict: For code generation, GPT-5.3 Codex ($648/mo) is purpose-built for code and costs 78% less than GPT-5.6 Sol. Claude Sonnet 5 ($540/mo) is also excellent for code at 82% less. DeepSeek V4 Pro ($96.60/mo) is the budget champion at 97% less.
Scenario 3: Document Analysis (100 documents/day)
Average: 15,000 input tokens, 1,000 output tokens per document. 30 days/month.
Monthly Document Analysis Cost
Verdict: For document analysis with large inputs, GPT-5.6's 1.05M context window is a genuine advantage — you can process larger documents without chunking. But Gemini 3.1 Pro ($750/mo) offers similar context at 76% less cost. For budget-conscious teams, Gemini 2.5 Flash-Lite ($57/mo) handles most document tasks at 98% less.
When GPT-5.6 Is Worth the Cost
- Massive context processing: The 1.05M context window is the largest available. If you need to process entire codebases, long legal documents, or multi-hour conversations in a single request, GPT-5.6 has no equal.
- Complex multi-step reasoning: GPT-5.6 Sol's reasoning quality justifies the premium for tasks where errors compound — financial analysis, legal review, medical documentation.
- OpenAI ecosystem integration: If you're already using OpenAI's API, GPT-5.6 Terra ($2/$12) offers a significant cost reduction from GPT-5.5 with the same quality tier.
- Budget-conscious premium: GPT-5.6 Luna ($0.20/$1.20) delivers GPT-5-level quality at 96% less cost than Sol — the best value in the GPT-5.6 family.
When GPT-5.6 Is Overkill
- Chatbots: Claude Sonnet 5 ($2/$10) or GPT-5.4 mini ($0.75/$4.50) handle 95% of chatbot queries at 60-96% less cost.
- Data extraction: GPT-4o mini ($0.15/$0.60) handles structured extraction at 97% less cost.
- Summarization: Gemini 2.5 Flash-Lite ($0.10/$0.40) handles summarization at 98% less cost.
- Classification: GPT-4o mini or DeepSeek V4 Flash handle classification tasks at 95%+ cost savings.
- High-volume simple tasks: DeepSeek V4 Flash ($0.22/$0.66) is the cheapest option for simple, high-volume workloads.
The Output Token Trap
Output tokens cost 6x more than input tokens for all GPT-5.6 models. This is the biggest hidden cost driver:
| Output Length | GPT-5.6 Sol | GPT-5.6 Terra | GPT-5.6 Luna |
|---|---|---|---|
| 200 tokens (short answer) | $0.006 | $0.003 | $0.0012 |
| 500 tokens (paragraph) | $0.015 | $0.0075 | $0.003 |
| 1,000 tokens (detailed response) | $0.030 | $0.015 | $0.006 |
| 4,000 tokens (long-form) | $0.120 | $0.060 | $0.024 |
Pro tip: Always set max_tokens. An unbounded GPT-5.6 Sol request generating 4,000 output tokens costs $0.12 in output alone — that's the cost of 800 GPT-4o mini requests.
How to Calculate Your GPT-5.6 Costs
Cost Formula
Monthly Cost = (Input Tokens × Price + Output Tokens × Price) × Requests per Month ÷ 1,000,000
Example with GPT-5.6 Terra: 200 requests/day × 3,000 input tokens × $2/1M + 200 × 1,200 output × $12/1M = $1.20/day input + $2.88/day output = $124/month
Or skip the math — use the APIpulse Cost Calculator to compare GPT-5.6 with Claude, Gemini, and DeepSeek side by side.
5 Ways to Reduce GPT-5.6 API Costs
- Start with GPT-5.6 Luna. At $0.20/$1.20, it's 96% cheaper than Sol with the same 1.05M context. Only upgrade to Terra or Sol if Luna's quality isn't sufficient for your use case.
- Implement model routing. Route simple queries to GPT-5.6 Luna ($0.20/$1.20), moderate tasks to Terra ($2/$12), and only complex reasoning to Sol ($4/$20). This can reduce costs 60-80%.
- Set max_tokens religiously. Output tokens cost 6x more than input. Setting max_tokens to 500 instead of leaving it unbounded can cut costs 50%.
- Use batch API for non-real-time workloads. OpenAI's batch API offers 50% discount on all models. For document processing, analysis, and other async tasks, this halves your GPT-5.6 costs.
- Consider Claude Sonnet 5. At $2/$10 (vs GPT-5.6 Terra's $2/$12), Claude Sonnet 5 is 20% cheaper on input and 33% cheaper on output with comparable quality.
The Bottom Line
GPT-5.6 Terra ($2/$12) is the sweet spot. It offers 60% cost reduction from Sol with the same 1.05M context window and comparable quality for most tasks. GPT-5.6 Luna ($0.20/$1.20) is the budget champion — 96% cheaper than Sol for high-volume workloads. GPT-5.6 Sol ($4/$20) is only worth it for the most demanding reasoning tasks where the 1.05M context window is essential. For everything else, you're overpaying.
Calculate your exact GPT-5.6 costs. Enter your usage and compare with every alternative.
Try the Free Cost Calculator or Compare All Models