Blended rate
Same for both
$0.20 per 1M each
Cheaper on long prompts
GPT-6 Luna
above 100,000 input tokens
Higher benchmark scores
Claude Haiku 5.5
leads on 5 of 8 figures
Bigger context window
GPT-6 Luna
1,050,000 vs 1,000,000 tokens
Where each one wins
Claude Haiku 5.5
- Higher overall intelligence (Intelligence Index) 43.4 vs 38.1
- Higher expert-exam performance (Humanity's Last Exam) 44.4 vs 38.5
- Higher scientific coding (SciCode) 55 vs 54.6
- Ahead on 2 more benchmark figures
GPT-6 Luna
- Cheaper on all four workloads
- Cheaper on long prompts above 100,000 input tokens
- Larger context window 1,050,000 tokens
- Higher long-context reasoning (AA-LCR) 83.3 vs 82.7
- Higher physics research reasoning (CritPt) 19.4 vs 18.9
- Higher factual accuracy (Omniscience Accuracy) 44.2 vs 36.4
- Offers Flex and Priority pricing not listed for Claude Haiku 5.5
What four workloads cost
One run, and the same run a thousand times.
They cost the same on a chat turn and a RAG answer and an agent loop, so the difference only shows up further down: on a long document, GPT-6 Luna costs $0.0823 against $0.2075.
| Workload | Claude Haiku 5.5 | GPT-6 Luna | ×1,000 | Difference |
|---|---|---|---|---|
| Chat turn 1,000 in / 500 out | $0.0004 | $0.0004 | $0.3500 / $0.3500 | Same |
| RAG answer 10,000 in / 800 out | $0.0014 | $0.0014 | $1.40 / $1.40 | Same |
| Agent loop 32,000 in / 700 out × 20 calls | $0.0170 | $0.0170 | $17.00 / $17.00 | Same |
| Long document 400,000 in / 3,000 out | $0.2075 | $0.0823 | $207.50 / $82.25 | GPT-6 Luna is 60% cheaper |
Price per 1M tokens
Above 100,000 input tokens the rates change, and the new rate applies to the whole request, not just the tokens past that point: Claude Haiku 5.5's input goes from $0.1000 to $0.5000. That is where the cheaper model changes.
| Context band | Price | Claude Haiku 5.5 | GPT-6 Luna | Difference |
|---|---|---|---|---|
| Prompts up to 100,000 tokens | Input | $0.10 | $0.10 | Same |
| Prompts up to 100,000 tokens | Cached input | $0.01 | $0.01 | Same |
| Prompts up to 100,000 tokens | Output | $0.50 | $0.50 | Same |
| Prompts up to 272,000 tokens | Input | $0.50 | $0.10 | GPT-6 Luna is 80% cheaper |
| Prompts up to 272,000 tokens | Cached input | $0.05 | $0.01 | GPT-6 Luna is 80% cheaper |
| Prompts up to 272,000 tokens | Output | $2.50 | $0.50 | GPT-6 Luna is 80% cheaper |
| Prompts over 272,000 tokens | Input | $0.50 | $0.20 | GPT-6 Luna is 60% cheaper |
| Prompts over 272,000 tokens | Cached input | $0.05 | $0.02 | GPT-6 Luna is 60% cheaper |
| Prompts over 272,000 tokens | Output | $2.50 | $0.75 | GPT-6 Luna is 70% cheaper |
Blended: Claude Haiku 5.5 $0.20, GPT-6 Luna $0.20 per 1M — one rate at 3:1 input to output, for comparing two models at a glance.
Claude Haiku 5.5 priced by Anthropic, GPT-6 Luna by OpenAI. Standard tier, pay-as-you-go.
Other pricing tiers
Both models offer Batch, worth up to 50% off their standard rates, and the cheaper model is the same one on every tier — so the tier you buy changes the bill rather than the choice. Only GPT-6 Luna lists Flex and Priority, at up to 50% off its own standard rate; Claude Haiku 5.5 offers none of them.
Every tier either model sells is here, so this is the whole price sheet. A saving is measured against that model's own standard rate — how we read tiers.
| Tier | Claude Haiku 5.5 | GPT-6 Luna | Cheaper |
|---|---|---|---|
| Batch | $0.05 in / $0.25 out 50% off Standard | $0.05 in / $0.25 out 50% off Standard | Same |
| Flex only on GPT-6 Luna | Not offered | $0.05 in / $0.25 out 50% off Standard | — |
| Priority only on GPT-6 Luna | Not offered | $0.20 in / $1.00 out | — |
Benchmarks
They share 8 figures: Claude Haiku 5.5 leads on 5, GPT-6 Luna on 3. The widest gap is 36.3 points, on answer reliability (Omniscience Non-Hallucination).
Overall intelligence
Intelligence Index — Overall intelligence across reasoning, knowledge & math evals
Claude Haiku 5.5 ahead by 5.3
Expert-exam performance
Humanity's Last Exam — Expert-level questions across many academic domains
Claude Haiku 5.5 ahead by 5.9
Long-context reasoning
AA-LCR — Long-context reasoning across large inputs
GPT-6 Luna ahead by 0.6
Scientific coding
SciCode — Research-level scientific coding tasks
Claude Haiku 5.5 ahead by 0.4
Physics research reasoning
CritPt — Unpublished physics research reasoning problems
GPT-6 Luna ahead by 0.5
Real-world task performance
GDPval — Economically valuable, real-world knowledge work
Claude Haiku 5.5 ahead by 9.5
Factual accuracy
Omniscience Accuracy — Breadth of factual knowledge across domains
GPT-6 Luna ahead by 7.8
Answer reliability
Omniscience Non-Hallucination — How reliably the model avoids fabricated answers
Claude Haiku 5.5 ahead by 36.3
Each figure is that model's best published run, with the effort level named beside it — where these scores come from.
Specifications
GPT-6 Luna holds 50,000 more input tokens in one request. Only GPT-6 Luna lists web search.
| Specification | Claude Haiku 5.5 | GPT-6 Luna |
|---|---|---|
| Context window | 1,000,000 tokens | 1,050,000 tokens |
| Max output | 128,000 tokens | 128,000 tokens |
| Takes in | Text, image, file | Text, image, file |
| Puts out | Text | Text |
| Knowledge cutoff | 30 Jun 2026 | 18 May 2026 |
| Released | 7 Oct 2026 | 22 Sep 2026 |
| Status | Active | Active |
| Sold by | Anthropic, Perplexity | OpenAI |
| Tool use | Yes | Yes |
| Structured outputs | Yes | Yes |
| Web search | Not listed | Yes |
| Prompt caching | Yes | Yes |
| Code execution | Yes | Yes |
| Computer use | Yes | Yes |
| Open weights | Not listed | Not listed |
| Reasoning | Yes | Yes |
FAQs
- Is Claude Haiku 5.5 cheaper than GPT-6 Luna?
- They cost the same on a typical exchange: 1,000 input tokens with a 500-token reply comes to $0.0004 on either.
- How much do Claude Haiku 5.5 and GPT-6 Luna cost per 1M tokens?
- Claude Haiku 5.5 costs $0.10 for input and $0.50 for output. GPT-6 Luna costs $0.10 and $0.50. Standard pay-as-you-go rates, as of 7 Oct 2026.
- Which is better, Claude Haiku 5.5 or GPT-6 Luna?
- On published benchmarks, Claude Haiku 5.5. It leads on 5 of the 8 figures both models report, including overall intelligence (Intelligence Index), where it scores 43.4 against 38.1.
- Is Claude Haiku 5.5 or GPT-6 Luna cheaper for long prompts?
- GPT-6 Luna, above 100,000 input tokens. A rate that changes at a context length applies to the whole request, not only the tokens past it, so the switch is sharp rather than gradual.
- Is Batch pricing cheaper for Claude Haiku 5.5 or GPT-6 Luna?
- Batch input costs $0.05 on Claude Haiku 5.5 and $0.05 on GPT-6 Luna, 50% below the standard rate on both.
- Does Claude Haiku 5.5 or GPT-6 Luna have a bigger context window?
- GPT-6 Luna, at 1,050,000 tokens against 1,000,000. That is 50,000 more input tokens in a single request.