Models, measured side by side
GPT-5.2 Pro vs Nemotron 3.5 Lightning
Compare GPT-5.2 Pro, Nemotron 3.5 Lightning by token pricing, context length, benchmark results, speed and tool support.
Monthly workload
Monthly cost breakdown
InputOutput
GPT-5.2 Pro
$1,260 / mo
Nemotron 3.5 Lightning
$1.68 / mo
Performance comparison
GPT-5.2 Pro
Not listed
Nemotron 3.5 Lightning
12.9
Overview
Pricing & capacity
| Pricing & capacity | GPT-5.2 Pro | Nemotron 3.5 Lightning |
|---|---|---|
| Input / 1M tokens | $21 | $0.05 |
| Output / 1M tokens | $168 | $0.14 |
| Cached input / 1M | Not listed | $0.02 |
| Monthly cost | $1,260 | $1.68 |
| Context window | 400K | 262K |
| Maximum output | 128K | 33K |
Input / 1M tokens
- GPT-5.2 Pro
- $21
- Nemotron 3.5 Lightning
- $0.05
Output / 1M tokens
- GPT-5.2 Pro
- $168
- Nemotron 3.5 Lightning
- $0.14
Cached input / 1M
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- $0.02
Monthly cost
- GPT-5.2 Pro
- $1,260
- Nemotron 3.5 Lightning
- $1.68
Context window
- GPT-5.2 Pro
- 400K
- Nemotron 3.5 Lightning
- 262K
Maximum output
- GPT-5.2 Pro
- 128K
- Nemotron 3.5 Lightning
- 33K
Benchmarks & performance
| Benchmarks & performance | GPT-5.2 Pro | Nemotron 3.5 Lightning |
|---|---|---|
| Arena Elo | 1,255 | Not listed |
| MMLU-Pro | 82.8% | Not listed |
| Intelligence Index | Not listed | 12.9 |
| GPQA | Not listed | 74.3% |
| Humanity’s Last Exam | Not listed | 10.6% |
| SciCode | Not listed | 32.1% |
| Output speed | Not listed | 301 tokens/s |
| Time to first token | Not listed | 0.56 s |
Arena Elo
- GPT-5.2 Pro
- 1,255
- Nemotron 3.5 Lightning
- Not listed
MMLU-Pro
- GPT-5.2 Pro
- 82.8%
- Nemotron 3.5 Lightning
- Not listed
Intelligence Index
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- 12.9
GPQA
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- 74.3%
Humanity’s Last Exam
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- 10.6%
SciCode
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- 32.1%
Output speed
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- 301 tokens/s
Time to first token
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- 0.56 s
Tools & features
| Tools & features | GPT-5.2 Pro | Nemotron 3.5 Lightning |
|---|---|---|
| Function calling | Supported | Supported |
| Structured outputs | Supported | Supported |
| JSON mode | Supported | Supported |
| Reasoning | Supported | Supported |
| Log probabilities | Not listed | Supported |
| Deterministic seed | Supported | Supported |
| Prompt caching | Not listed | Supported |
Function calling
- GPT-5.2 Pro
- Supported
- Nemotron 3.5 Lightning
- Supported
Structured outputs
- GPT-5.2 Pro
- Supported
- Nemotron 3.5 Lightning
- Supported
JSON mode
- GPT-5.2 Pro
- Supported
- Nemotron 3.5 Lightning
- Supported
Reasoning
- GPT-5.2 Pro
- Supported
- Nemotron 3.5 Lightning
- Supported
Log probabilities
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- Supported
Deterministic seed
- GPT-5.2 Pro
- Supported
- Nemotron 3.5 Lightning
- Supported
Prompt caching
- GPT-5.2 Pro
- Not listed
- Nemotron 3.5 Lightning
- Supported
API & availability
| API & availability | GPT-5.2 Pro | Nemotron 3.5 Lightning |
|---|---|---|
| API identifier | gpt-5.2-pro | nemotron-3.5-lightning |
| API providers | OpenAI | Not listed |
| Prices checked | Oct 8, 2026 | Oct 8, 2026 |
API identifier
- GPT-5.2 Pro
gpt-5.2-pro- Nemotron 3.5 Lightning
nemotron-3.5-lightning
API providers
- GPT-5.2 Pro
- OpenAI
- Nemotron 3.5 Lightning
- Not listed
Prices checked
- GPT-5.2 Pro
- Oct 8, 2026
- Nemotron 3.5 Lightning
- Oct 8, 2026
About the models
GPT-5.2 Pro
GPT-5.2 with extra compute for the most demanding problems. No cached-input discount.
Full pricing & detailsNemotron 3.5 Lightning
NVIDIA Nemotron 3.5 Lightning is an open mixture-of-experts model from NVIDIA, with 3B active parameters out of 30B total. It is suited for high-throughput agentic workloads and specialized tasks that...
Full pricing & detailsComparison FAQ
GPT-5.2 Pro: $21 input and $168 output per million tokens. Nemotron 3.5 Lightning: $0.05 input and $0.14 output per million tokens. The cheaper choice depends on your input-to-output ratio; a model can have a lower input rate but a higher output rate.