OpenAIOpenAI·text

GPT-5.2

ReasoningWeb searchFunction calling
Advertised as gpt-5.2gpt-52openai-gpt-52
Quick reference
GPT-5.2 — TLDR
  • 🆕 OpenAI's GPT-5 series frontier model, released December 2025.
  • 🧠 Adaptive reasoning allocates compute dynamically across simple and complex tasks.
  • 📏 256K-token context window for long-document and long-horizon work.
  • 🔧 Built for agentic workflows with stronger, more reliable tool calling.
  • 🎯 Reasoning effort tunable from none through extra-high.
  • 👁️ Adds reasoning, function-calling, and web-search capabilities.
  • ⚡ Token-efficient reasoning lowers cost-to-quality on agentic evals.
  • 🏢 Positioned by OpenAI for professional knowledge work and long-running agents.
💰 Best price on AntSeed
$0.030 / $0.133
per 1M · cheapest in / out
📏 Context
256K tokens
🐜 Sellers
9
advertising on AntSeed
Provider

OpenAI is an American artificial intelligence research organization headquartered in San Francisco, structured as both a for-profit public benefit corporation and a nonprofit foundation. The lab developed the GPT family of large language models, the DALL-E image generation…

Explore 27 more models by OpenAI
About this model

GPT-5.2 is a frontier-grade model in OpenAI's GPT-5 series, released in December 2025 and described by OpenAI as its most advanced model for everyday professional work, spanning reasoning, long-context understanding, coding, and vision. It uses adaptive reasoning to allocate computation dynamically, answering simple queries quickly while spending more depth on complex tasks, and supports a 256K-token context window alongside function-calling and web-search.

Compared with its predecessor GPT-5.1, OpenAI reports stronger multi-step reasoning, greater quantitative accuracy, and more reliable problem solving on complex technical tasks. A key theme is token efficiency: OpenAI states that despite a higher per-token price, the cost of reaching a given quality level can end up lower because GPT-5.2 uses fewer tokens, and the company highlights improved tool calling and lower latency in agentic workflows. In the API, reasoning effort is configurable from none through extra-high.

Within the GPT-5 family, GPT-5.2 has a coding-specialized counterpart in GPT-5.2 Codex, which OpenAI tuned for long-horizon agentic coding, context compaction, large refactors, and Windows environments. It was later succeeded by GPT-5.4 and GPT-5.5, with OpenAI's documentation now recommending the newer GPT-5.5 for the most complex professional work and listing GPT-5.2 as a previous frontier model.

For teams, GPT-5.2 targets long-running agents, grounded assistants, and tool-heavy production workflows, building on the GPT-5 lineage with an emphasis on reliability and efficient reasoning rather than raw scale.

Sources
openai.comIntroducing GPT-5.2 | OpenAI· openai.complatform.openai.comGPT-5.2 Model | OpenAI API· platform.openai.comdevelopers.openai.comGPT-5.5 Model | OpenAI API· developers.openai.com

This About section is AI-generated from public sources via VeniceStats + Venice inference, with no human editing. It may contain inaccuracies.

Usage on AntSeed
Tokens served
243.45k
input + output
Requests
32
settled calls
Buyers
6
distinct, on this model
Sellers used
5
of 9 advertising
Settled
$0.32
gross USDC, this model
Sellers serving GPT-5.2 (9)compare on the network explorer →
SellerReputationRoutingInput $/MCached $/MOutput $/MCategoriesAPI

"Best price" and the seller table are live AntSeed catalog data (advertised $/1M tokens — or $ per generated image for unit-billed image models — not settled amounts). Reputation = on-chain trust (0-100). "Routing" = the SDK's default buyer routing (what the VPR desktop app ships with): a trust ≥ 60 gate on the effective reputation, then cheapest-first among routable sellers; live failover state (per-peer cooldowns) is buyer-side runtime and not included. Model knowledge (TLDR, provider, About) via the VeniceStats enrichment layer. Advertised catalog, not the model used in any specific purchase. "Usage on AntSeed" counts only settlements whose buyers share the per-model split on-chain (metadata v2/v3, opt-in), so every usage figure is a lower bound.