GPT-5.2
OpenAIProprietaryReasoningVerifiedReleased Dec 11, 2025Input: file, image, text
Overall score
59.9
Rank 57 of 222 · 52.8–67.0
Context window
400K
400,000 tokens
Price per 1M tokens
$1.75 in · $14.00 out
✓ Verified on OpenAI's pricing page
Blended price
$4.81
3:1 input:output tokens
Output speed
—
Not measured yet
Category scores
Scores per capability area with the model's rank among ranked models in that category.
- 1Instruction following#1992.6
- 2Multimodal#2275.9
- 3Knowledge#5175.9
- 4Math71.1
- 5Reasoning64.3
- 6Coding#7551.4
- 7Agentic & tool use26.9
Benchmark results (31)
As reported in system cards, papers and public evaluations. Position compares this model with every other model reporting the same benchmark.
Multimodal
| Design Arena Website | 1196 | #56 of 91 |
| MMMU-Pro | 79.5 | #12 of 48 |
| CharXiv | 82.1 | #19 of 42 |
| MMMU | 86.7 | #2 of 21 |
| MathVision | 83.0 | #17 of 20 |
| SimpleVQA | 55.8 | #16 of 17 |
| RealWorldQA | 83.3 | #9 of 16 |
| ERQA | 59.8 | #13 of 16 |
| VideoMMMU | 85.9 | #5 of 12 |
| V* | 75.9 | #11 of 12 |
| OmniDocBench 1.5 | 85.7 | #10 of 11 |
| Video-MME (with subtitle) | 86.0 | #9 of 11 |
| MLVU (M-Avg) | 85.6 | #6 of 9 |
| CountBench | 91.9 | #5 of 8 |
| Video-MME (w/o subtitle) | 85.8 | #2 of 7 |
| CC-OCR | 70.3 | #7 of 7 |
| AI2D_TEST | 92.2 | #4 of 7 |
| DynaMath | 86.8 | #1 of 6 |
| We-Math | 79.0 | #4 of 5 |
Agentic & tool use
| Gert Labs | 46.54 | #28 of 52 |
| BrowseComp | 65.8 | #36 of 50 |
| OSWorld-Verified | 47.3 | #34 of 35 |
| JobBench | 34.3 | #20 of 34 |
Math
| FrontierMath v2 (Tiers 1-3) | 67.400 | #26 of 64 |
| FrontierMath v2 (Tier 4) | 31.700 | #26 of 51 |
| AIME 2025 | 100.0 | #1 of 38 |
Coding
| SWE-bench Verified | 80.0 | #19 of 98 |
| SWE-bench Pro | 55.6 | #46 of 69 |
Reasoning
| ARC-AGI-2 | 52.9 | #33 of 41 |
| ARC-AGI-1 | 86.17 | #28 of 36 |
Knowledge
| GPQA | 92.4 | #12 of 117 |
Pricing by source
USD per 1M tokens.
| Source | Input | Output | Cached in |
|---|---|---|---|
| ✓ Official (OpenAI) checked Oct 17, 2026 · gpt-5.2 | $1.75 | $14.00 | $0.17 |
| BenchLM list price | $1.75 | $14.00 | $0.17 |
| OpenRouter openai/gpt-5.2 | $1.75 | $14.00 | $0.17 |
Standard tier, prompts under 272K tokens.
OpenAI's GPT-5.2 model page lists GPT-5.2 at $1.75 input / $14.00 output per million tokens.
Arena ratings
Human-preference ratings from LMArena. Best-rated variant shown.
| Text(high) | 1416 | #129 · 49.7K votes |
| WebDev (coding) | 1417 | #89 · 1.1K votes |
| Vision(High) | 1246 | #68 · 19K votes |
| Documents(high) | 1405 | #41 · 7.1K votes |
Speed
Measured by us through OpenRouter (median of trials).
Not measured yet.
Related models
Same family
- GPT-5.2 Pro66.7
Similar score
- Qwen3.8-27B59.6 · $0.37
- Muse Spark60.5 · —
- GPT-5.3 Codex60.5 · $4.81
- GLM-5.260.5 · $2.15
- Gemini 3 Pro61.2 · $4.50