Enterprise Business Development Manager

OpenAI Web Search scores 74 but loses on cost to cheaper rivals. OpenAI's built-in web_search tool lands fifth on the Artificial Analysis Search Index at 74 points, trailing Perplexity, Octen, Parallel, and Brave. · 3 hrs ago Read 5 min * OpenAI Web Search debuts at 74 on the Artificial Analysis Search Index, ranking 5th among providers. * Costs around $0.05 per task: $0.04 search ($10 per 1k calls) plus $0.009 model inference. * Scores 72% on AA-Omniscience, nearly matching leader Firecrawl at 73%. * Weakest on BrowseComp multi-hop browsing at 73.5%, 13th of 26 variants. * Uses ~40k input tokens per task versus 125k for the leanest external Search API. * Perplexity (medium) leads at 80, Octen beats OpenAI at half the price. OpenAI Web Search scores 74 in its first Search API benchmark. Artificial Analysis has added OpenAI Web Search to its Search Index, giving the built-in tool its first direct comparison with standalone Search APIs. OpenAI scored 74, ranking fifth among providers and seventh among 26 product configurations. The result shows strong factual retrieval and efficient token use, alongside weaker performance on research that requires several linked searches. One call, different plumbing. OpenAI Web Search is the index's first integrated, first-party search tool. Standalone providers use Stirrup, an agent harness that lets a candidate model call an external search endpoint, inspect extracted pages, and repeat the process. OpenAI handles retrieval, reasoning, and answer generation inside one Responses API request, with the model choosing when and how often to search. Artificial Analysis paired each provider with GPT-5.6 Luna at medium reasoning effort. The standard setup allowed up to 25 agent turns, returned as many as 10 results per search, and extracted page text with a 15-second timeout. The OpenAI run bypassed Stirrup and set search_context_size to medium. The common model makes answer quality more comparable across providers, although the orchestration remains different. OpenAI's score reflects its complete integrated workflow, while standalone API scores include Stirrup's query planning, page extraction, and result handling. Factual lookup leads its profile. The Search Index gives equal weight to three benchmarks. Its published score uses the underlying results, while the component figures below are displayed at lower precision. OpenAI Web Search benchmark results | Benchmark | Task | Result | Relative position | | AA-Omniscience | Factual question answering | 72% accuracy | Near Firecrawl's leading 73% | | DeepSearchQA | Broad research | 78 F1 | Mid-pack | | BrowseComp | Multi-hop web browsing | 74% accuracy | 13th of 26 | F1 combines precision and recall into one score, rewarding answers that recover relevant information without adding excessive incorrect material. BrowseComp measures a different capability: following clues across multiple searches and pages to reach an exact answer. Artificial Analysis used a 200-question subset drawn from BrowseComp's 1,266-question evaluation pool. OpenAI's strongest relative result came from direct factual questions. Its 74% BrowseComp accuracy trailed Perplexity and Octen, which scored between 85% and 87%. That 11-to-13-point gap matters for agents that investigate obscure entities, reconcile details across sites, or build answers through several search steps. GPT-5.6 Luna scored 33 without search, so the integrated tool added 41 index points. Perplexity's medium configuration produced a 47-point gain with the same candidate model. A score of 74 costs about five cents. OpenAI charges $10 per 1,000 web_search calls and bills retrieved search content as model input tokens. Under the benchmark workload, search averaged $40.19 per 1,000 tasks and model inference added $9.40. The combined cost was $49.59 per 1,000 tasks, or about $0.05 each. Quality, cost, and latency for selected configurations | Provider | Index score | Search per 1,000 tasks | Model per 1,000 tasks | Total per 1,000 tasks | Time per task | | Perplexity, medium | 80 | $62.30 | $8.04 | $70.34 | 27.5 seconds | | Octen, highlights | 77 | $9.07 | $14.92 | $23.99 | 15.9 seconds | | Parallel, advanced | 75 | $47.93 | $11.29 | $59.22 | 41.4 seconds | | Brave, LLM context | 75 | $61.96 | $17.46 | $79.42 | 22.7 seconds | | OpenAI Web Search, medium | 74 | $40.19 | $9.40 | $49.59 | 30.4 seconds | | Model only | 33 | N/A | $2.92 | $2.92 | 21.0 seconds | OpenAI was cheaper than 17 of the 25 standalone Search APIs in the index. Octen's API delivered a higher score for less than half the total cost, placing OpenAI outside the quality-cost frontier in this run. Perplexity led on quality at a higher total price. Integrated retrieval cuts the payload. OpenAI consumed roughly 40,000 input tokens per task, including retrieved content. The smallest measured footprint among the external Search API configurations was about 125,000 tokens. That difference kept OpenAI's model-inference cost near $0.009 per task, even with GPT-5.6 Luna performing the reasoning. Smaller context payloads can reduce input-processing latency, memory use for the key-value cache that stores prior token states, and exposure to context-based pricing. Those savings become more relevant when an application launches searches in parallel or runs repeated research loops. The API surface stays small. A Responses API request gains web access through a tool declaration. The model can then decide whether to search and can issue several searches before returning an answer. from openai import OpenAI client = OpenAI response = client.responses.create( model="gpt-5.6-luna", reasoning={"effort": "medium"}, tools=[{ "type": "web_search", "search_context_size": "medium"}], input="Your question here") print(response.output_text) One Responses request can trigger multiple billable search calls. The benchmark's $40.19 search cost per 1,000 tasks, combined with OpenAI's $10 rate per 1,000 calls, corresponds to roughly four search calls per task on average. OpenAI does not expose search duration separately, so Artificial Analysis estimates it from streamed events. The reported search segment is an upper bound that includes some OpenAI processing time, which limits direct latency comparisons with standalone APIs. The benchmark also omits per-query metrics because an integrated request can generate several internal queries. Where the built-in tool fits. Teams already using the Responses API can remove a separate search integration, including vendor authentication, request formatting, result parsing, and some retry handling. The benchmark supports the built-in tool for workloads with these characteristics: * Questions center on direct factual lookup. * Low token use matters for cost, latency, or context limits. * A single API and billing relationship simplifies operations. * End-to-end answer generation matters more than direct control over raw search results. BrowseComp-style research provides a strong reason to test external providers. Perplexity Search and Octen led OpenAI by more than 10 points on that benchmark, and Octen paired its higher overall score with substantially lower cost. Production evaluations should also measure source coverage, freshness, citation accuracy, regional behavior, rate limits, failure handling, and data-retention terms. The Artificial Analysis index captures answer quality, cost, token use, and latency under one controlled configuration; application-specific tests remain necessary before selecting a search stack.

near.com's Confidential TVL reaches $70 million, triggering the [email protected] snapshot amid industry-wide privacy wave. Sep 17, 2026, 10:00 ET The milestone signals a new era of confidential DeFi, providing protocols, institutions, and everyday users with a proven privacy layer for cross-chain transactions SAN FRANCISCO, Sept. 17, 2026 /PRNewswire/ - near.com - the app that brings crypto, tokenized assets, earning, trading, and payments into one connected experience, confidential by default - today announced that it has surpassed $70 million in confidential Total Value Locked (TVL). Hitting the $70M mark automatically triggers the snapshot for "Drop 1" in the [email protected] Incentive Milestone Program. 333,333 milestone tokens will soon be available for eligible community members to claim. Surpassing $70 million in Confidential TVL signals that confidential execution is becoming essential infrastructure for onchain markets. Confidential Intents routes transactions through a NEAR private shard, eliminating mempool exposure, predatory MEV, and front-running without the computational overhead of zero-knowledge systems. This growing pool of confidential liquidity gives institutions, enterprises, and autonomous AI agents the privacy and depth to execute high-volume cross-chain swaps without anyone tracing them back to who's behind it. Reaching the $70 million milestone has triggered the snapshot for the first 333,333 token reward, which will be available to claim through near.com. Rather than a traditional airdrop, the program is designed to reward users who stay active. Eligible participants must maintain a confidential balance above $100 and an active swap history, while rewards remain locked until NEAR's three-day VWAP reaches $3.33 or higher, at which point milestone tokens become NEAR. A 2% per-wallet cap also helps limit concentration and keep the distribution broad. Alex Shevchenko, GM of NEAR Intents said, "Confidentiality is quickly becoming a core requirement for the industry, and NEAR is becoming the rail to make it the new default. This milestone shows that confidential execution is moving beyond a niche use case and becoming the foundation for the next generation of onchain finance. By combining cross-chain liquidity with protection against front-running, strategy leakage, and other forms of MEV, NEAR is giving institutions and DeFi users a more secure way to transact at scale while preserving the speed and usability they expect." This snapshot marks the end of the first phase of the [email protected] campaign. NEAR currently powers confidential execution across more than 30 connected blockchains. Live TVL metrics are available at https://dune.com/near/near-intents. Incentive program details and eligibility criteria are available at https://www.near.org/blog/milestone-incentive-program. near.com is the only place to claim milestone tokens. Terms apply. About NEAR: NEAR is the open infrastructure powering the agent economy. NEAR's vertically integrated stack connects NEAR Intents for cross-chain execution, NEAR AI for confidential and verifiable AI, and sharded blockchain infrastructure for 1M+ TPS scalability. near.com is NEAR's unified front-end: one account, 30+ chains, and confidential cross-chain swaps, perps, RWAs, and earn. With $30B+ in cross-chain volume and AI integrations with Abound, Brave Nightly, the Government of Bermuda, OpenMind, Venice AI, and more, NEAR is where agents transact. Learn more at https://www.near.org/. SOURCE NEAR

Brave Wallet on desktop adds Polkadot and private cross-chain transfers in v1.94. Brave has added two wallet upgrades in the latest v1.94 version on desktop. It includes native Polkadot support and private cross-chain transfers powered by NEAR Protocol's "Confidential Intents." Brave officially posted this information on X. You can view the entire thread attached below. This is a desktop Brave Wallet update, covering Windows, macOS, and Linux; it is not an iOS or Android wallet announcement. Users already on Brave desktop can pick it up by opening the browser menu, heading over to About Brave, and installing the update. Since the wallet lives inside the browser, no separate app is required. The first change is Polkadot. You can now buy, hold, and send DOT directly in Brave Wallet. The wallet also supports AssetHub, the part of Polkadot used to issue and manage Polkadot-native assets. Furthermore, Swaps and broader Web3 connectivity for Polkadot are planned later this year, and are not included in this release. For those unaware, Polkadot is a network built so that many blockchains can share security and move assets between them. DOT is its native token, and AssetHub is the chain where those native tokens are created and managed. In simpler terms, Brave users no longer need a separate Polkadot wallet just to hold or send DOT. The second change is Confidential Intents. Brave says you can now move funds on Bitcoin, Cardano, Ethereum, and Solana without leaving the usual public trail on those chains, and Zcash support is coming later. By default, those transactions are routed through a NEAR private shard. More information about Confidential Intents is available here, but it's basically a privacy layer. Instead of broadcasting swap details on a public blockchain, the transfer runs on a private layer that you can't browse publicly. The size, pair, and path of the trade aren't as easily visible as a public transaction. While the incoming & outgoing money is still visible on public networks, the entire process in the middle is obscured. In other Brave news, the Containers feature is getting an upgrade, with upcoming automation features. More on that here. PiunikaWeb stand out from the tech-media crowd because PiunikaWeb break news stories; PiunikaWeb mainly bring you stuff that you won't find anywhere in the mainstream tech media. Its stories have been picked up by some of the world's most popular websites and media outlets - more info is available here. I have been a consumer technology enthusiast for over 5 years. Thanks to my experience in software beta testing and product reviews, I've understood and learnt a lot about what bugs and issues bother people, and I spend time trying to simplify their solutions. I cover smartphones, software, social media, apps, AI, and most consumer tech gadgets. Actively pursuing a Computer Science bachelor's degree. I'm mostly active on Twitter/X (@TechWhirlUlt), drop a DM or tag me if you want to share info or connect!