GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly to simple queries while spending more depth on complex tasks.
Built for broad task coverage, GPT-5.2 delivers consistent gains across math, coding, sciende, and tool calling workloads, with more coherent long-form answers and improved tool-use reliability.
Providers
Different companies host the same model. OpenRouter routes your request to one of them based on the routing mode you pick: Balanced (price + speed), Nitro (fastest), or Floor (cheapest).
BYOK only | $0.875 | $7.00 | $0.0875 | -- | -- |
|---|
Input $0.875 / Output $7.00-- tok/s-- uptimeBYOK only
Pricing
The average price customers actually pay for this model, next to the prices providers post. Caching and discounts mean the price actually paid is often well below the listed one.
GPT-5.2 (batch) — Price history
| $0.875 | $7.00 | $0.0875 | — |
- Input $0.875 / Output $7.00
Performance
Throughput is how fast the model writes (tokens per second — higher is better). Latency is total round-trip time (lower is better). TTFT is time-to-first-token — how long before you see anything appear (lower is better).
Uptime
Uptime is the percentage of the past 3 days that at least one provider was responding to requests. Availability is the percentage of time that inference was successfully served. OpenRouter continuously monitors and uses the next-best provider when one returns an error.
Not enough uptime data to display yet.
When an error occurs in an upstream provider, we can recover by routing to another healthy provider, if your request filters allow it. You can access per-provider uptime data programmatically through the Endpoints API. Learn more about our load balancing and provider routing options.
Benchmarks
Scores on standardized evaluations and this model’s own outputs on shared benchmark prompts. Higher percentages are better — and each rank shows where this model lands among the models evaluated on that source.
| Source | Benchmark | Score |
|---|---|---|
| Artificial Analysis | GPT-5.2 (Non-reasoning) GPQA Diamond | 71.2% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) HLE | 8.0% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) IFBench | 47.4% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) τ²-Bench Telecom | 46.5% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) AA-LCR | 45.7% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) τ-Bench Banking | 11.1% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) CritPt | 0.6% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) LiveCodeBench | 66.9% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) Terminal-Bench Hard | 31.8% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) MMLU-Pro | 81.4% |
| Artificial Analysis | GPT-5.2 (Non-reasoning) AIME 2025 | 51.0% |
| Artificial Analysis | GPT-5.2 (Medium) GPQA Diamond | 86.4% |
| Artificial Analysis | GPT-5.2 (Medium) HLE | 26.7% |
| Artificial Analysis | GPT-5.2 (Medium) IFBench | 65.2% |
| Artificial Analysis | GPT-5.2 (Medium) τ²-Bench Telecom | 74.3% |
| Artificial Analysis | GPT-5.2 (Medium) AA-LCR | 70.3% |
| Artificial Analysis | GPT-5.2 (Medium) CritPt | 7.9% |
| Artificial Analysis | GPT-5.2 (Medium) LiveCodeBench | 89.4% |
| Artificial Analysis | GPT-5.2 (Medium) Terminal-Bench Hard | 43.2% |
| Artificial Analysis | GPT-5.2 (Medium) MMLU-Pro | 85.9% |
| Artificial Analysis | GPT-5.2 (Medium) AIME 2025 | 96.7% |
| Artificial Analysis | GPT-5.2 (Xhigh) GPQA Diamond | 90.3% |
| Artificial Analysis | GPT-5.2 (Xhigh) HLE | 37.7% |
| Artificial Analysis | GPT-5.2 (Xhigh) IFBench | 75.4% |
| Artificial Analysis | GPT-5.2 (Xhigh) τ²-Bench Telecom | 84.8% |
| Artificial Analysis | GPT-5.2 (Xhigh) AA-LCR | 82.7% |
| Artificial Analysis | GPT-5.2 (Xhigh) CritPt | 11.6% |
| Artificial Analysis | GPT-5.2 (Xhigh) LiveCodeBench | 88.9% |
| Artificial Analysis | GPT-5.2 (Xhigh) Terminal-Bench Hard | 47.0% |
| Artificial Analysis | GPT-5.2 (Xhigh) MMLU-Pro | 87.4% |
| Artificial Analysis | GPT-5.2 (Xhigh) AIME 2025 | 99.0% |
| Design Arena | GPT-5.2 (Low) Models Arena 3D Elo | 1047 |
| Design Arena | GPT-5.2 (Low) Models Arena Asciiart Elo | 1147 |
| Design Arena | GPT-5.2 (Low) Models Arena Code Categories Elo | 1155 |
| Design Arena | GPT-5.2 (Low) Models Arena Data Visualization Elo | 1183 |
| Design Arena | GPT-5.2 (Low) Models Arena Game Development Elo | 1184 |
| Design Arena | GPT-5.2 (Low) Models Arena SVG Elo | 1125 |
| Design Arena | GPT-5.2 (Low) Models Arena UI Component Elo | 1149 |
| Design Arena | GPT-5.2 (Low) Models Arena Website Elo | 1168 |
| Vals AI | GPT 5.2 AIME | 96.9% |
| Vals AI | GPT 5.2 CaseLaw v2 | 66.0% |
| Vals AI | GPT 5.2 CorpFin v2 | 65.9% |
| Vals AI | GPT 5.2 Finance Agent v1.1 | 58.5% |
| Vals AI | GPT 5.2 GPQA Diamond | 91.7% |
| Vals AI | GPT 5.2 IOI v1 | 54.8% |
| Vals AI | GPT 5.2 LiveCodeBench | 85.4% |
| Vals AI | GPT 5.2 LegalBench | 82.8% |
| Vals AI | GPT 5.2 MedCode | 49.7% |
| Vals AI | GPT 5.2 MedQA | 94.1% |
| Vals AI | GPT 5.2 MedScribe | 84.4% |
| Vals AI | GPT 5.2 MGSM | 94.0% |
| Vals AI | GPT 5.2 MMLU Pro | 86.2% |
| Vals AI | GPT 5.2 MMMU Pro | 86.7% |
| Vals AI | GPT 5.2 MortgageTax | 67.1% |
| Vals AI | GPT 5.2 ProofBench v1 | 15.0% |
| Vals AI | GPT 5.2 Vals RSI Index | 15.8% |
| Vals AI | GPT 5.2 SAGE v1 | 49.3% |
| Vals AI | GPT 5.2 SWE-bench | 75.8% |
| Vals AI | GPT 5.2 TaxEval v2 | 75.8% |
| Vals AI | GPT 5.2 Terminal-Bench 2.0 | 51.7% |
| Vals AI | GPT 5.2 Vals Index | 64.5% |
| Vals AI | GPT 5.2 Vals Multimodal Index | 63.7% |
| Vals AI | GPT 5.2 Vibe Code Bench v1.1 | 53.5% |
Apps
Public apps that send the most traffic to this model. Good signal for what real production workloads look like — and a hint at which use cases this model is best suited for.
Not enough data for apps using this model yet.
Activity
Token volume and request traffic to this model over time.
API
Drop-in code to call this model. OpenRouter's API is OpenAI-compatible — most SDKs work by just swapping the base URL. The only thing that changes between models is the model slug below.
Not currently available through EU in-region routing. In-region routing docs
Frequently asked questions
GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1. It uses adaptive reasoning to allocate computation dynamically, responding quickly to simple queries while spending more depth on complex tasks.
GPT-5.2 (batch) costs $0.875/M input tokens and $7.00/M output tokens, with separate rates for Cache Read at $0.0875/M tokens and Web Search at $10.00/1K calls.
GPT-5.2 (batch) has a 400,000 token context window. It supports up to 128,000 completion tokens.
Yes. GPT-5.2 (batch) accepts tools and tool_choice for function calling. It also supports structured outputs via a JSON schema in response_format.
GPT-5.2 (batch) accepts files such as PDFs, images, and text as input and returns text.
GPT-5.2 (batch) is served by 1 provider on OpenRouter: OpenAI.
GPT-6.1 Terra Pro, GPT-6.1 Terra, GPT-6 Luna Decisions and 58 more are other text models from OpenAI.
GPT-5.2 (batch) was released on December 10, 2025.