Source: models: pplx-embed-v2-context and pplx-embed-v2-late; sizes 0.6B and 9B; shared embedding space enabling cross-model querying; publicly available on Hugging Face; evaluation on ViDoRe (V3), Q2D-Web, Browse, MADQA; agentic retrieval; dates: release of pplx-embed-v2-context recently; embedding models released…
The page documents a material pricing change for browserfoundry deep search: $3.00 per 1,000 requests, with live full-page reads and re-ranking, no model in the loop, as part of a now-focused agent-oriented web search pipeline. Source: browserfoundry blog benchmark page.
The page discusses Exa’s product changes: Exa retired the "/research" endpoint in April 2026 in favor of "/search" with type: "deep-reasoning", and marks "/findSimilar" as deprecated in the OpenAPI spec, directing developers to describe the source page in a "/search" query. It also mentions pricing and comparisons…
The page describes Perplexity changing the default web retrieval mode in its Agent API to Fast Search for low/medium/high presets, with web search pricing reducing from $2.50 to $1 per 1,000 calls and ~800 ms latency improvement. The xhigh preset remains standard web search; setting search_type: "web" preserves…
You.com is now live as a named connector in Merge Agent Handler with nine tools exposed (search_web, get_contents, get_answer, run_research, get_research_task, list_research_events, run_finance_research, search_images, get_account_balance) as part of the MCP server, plus details on authentication, credential storage,…
URL: (article reports Perplexity releasing two new multimodal search models (pplx-embed-v2-late) with late interaction, 9B and 0.6B, enabling image-based PDF search without OCR; indexing with 9B offline, querying with 0.6B online; storage per word piece 128 numbers; benchmarks and availability on Hugging Face; no…
AI Search is now generally available | Cloudflare Blog.
Official Gemini API changelog announces deprecation of gemini-3.7-flash and gemini-3.5-flash, automatic routing to gemini-3.8-flash and gemini-3.6-flash, with a final shutdown date for deep-research-pro-preview-12-2025 on 2026-10-23; introductory pricing for Gemini 3.8 Flash through 2026-12-31: $0.75 per 1M input…
URL: This page announces the public launch of GPT-6 Sol and GPT-6 Luna across ChatGPT, replacing GPT-5.6 Sol/Luna, with distinctions for October releases (vs September in Codex/Work). It references safety evaluations and capabilities, including High capability in Biological/Chemical and Cybersecurity domains, and…
The page documents multiple Perplexity API release notes dated Oct 7, 2026, including Claude Haiku 5.5 The Agent API with anthropic/claude-haiku-5-5, and various related API pricing and capability updates; no explicit model IDs/pricing changes beyond what's listed in the text. The page serves as a changelog for…
OpenAI Web Search debuts in the Artificial Analysis Search Index with a score of 74, ranking 5th among providers behind Perplexity, Octen, Parallel and Brave. The configuration uses GPT-5.6 Luna with medium reasoning and the built-in web_search tool; pricing cited as about $0.05 per task for the built-in search,…
— Perplexity’s Agent API presets now use Fast Search for web_search, reducing calls from $2.50 to $1.00 per 1,000 invocations (xhigh preset unchanged)
Includes material changes such as Gemini 2.5 models access update (restricting to users with prior usage), antigravity agent deprecation/replace, Interactions API launch, File Search API public preview launch, URL context tool GA with deprecation timeline, Deep Research agent previews and capabilities, File input…
Prices range from $1 to $14 per 1,000 calls for standard modes, with specific rates: Parallel and Perplexity fast modes at $1; Brave, Perplexity standard, Linkup basic, Parallel basic at $5; Exa default auto at $7; Tavily basic search at $8; OpenAI search tools at $10; Gemini grounding Google Search at $14 after…
Pricing and modes described: Turbo ($0.001 per request, ~200ms), Fast ($0.001 per request, ~1s), Basic ($0.005 per request, ~1s), Advanced ($0.005 per request, ~3s) with 10 results included; free tier provides $5 in credits monthly (up to 5,000 Turbo requests). Independent benchmarks (September 2026) compare Parallel…
Cloudflare announces the Web Search API via AI Gateway with ground truthing through partnerships with Ceramic.ai, Exa, and Linkup; includes Bring-Your-Own-Key (BYOK), REST endpoints, and integration into AI Gateway with observability, billing, and access controls; pricing available via partner web search API pricing;…
Page documents multiple October 2026 AI API price changes across OpenAI, Perplexity, Google Gemini, xAI, Anthropic, Meta/DeepInfra, and DeepSeek, including GPT-6.1 Sol and GPT-6 Luna launches with new price points; Perplexity Sonar migrated to Agent API with updated pricing; Gemini 2.5 limited to existing developers;…
URL: — Changelog entry: AI Search is generally available; includes usage-based billing begins on 2026-11-01 with included usage; default hybrid search; Workers AI embeddings and reranking included; multimodal model support; OCR availability and file size limits; source type inference.
— Google announces Gemini 4 Argon frontier model, with Fairwind Program rollout, pricing: introductory price 2 USD per million input tokens and 10 USD per million output tokens (cached input tokens 95% off), output token limit expanded to 1,000,000 tokens, and security/safety features roadmap; public rollout to…
The page documents multiple material changes across AI deep-research tooling in 2026, including: retirement of OpenAI's legacy Deep Research mode on March 26, 2026; replacement by an agent using the latest model by default (per OpenAI Deep Research API guide); production agent reportedly moving to GPT-5.6 Sol-class…
Embed 5 is generally available today on the Cohere API and Model Vault, Microsoft Foundry, and Amazon SageMaker. Pricing is $0.12 per million tokens for Pro and $0.08 per million tokens for Fast.
The page announces Gemini 4 Argon, a frontier model from Google Gemini, with rollout to trusted cyber defenders, a phased release without cyber guardrails for defenders, and initial pricing: $2 per million input tokens and $10 per million output tokens; later price change to $4 per 1M input tokens and $20 per 1M…
Page documents deprecation of Sonar models (sonar, sonar-deep-research, sonar-pro, sonar-reasoning-pro) and migration to Perplexity Agent API, including a live table of Sonar models with inputs USD / 1M and outputs USD / 1M, context tokens, and statuses; states that Sonar Chat Completions ended on 2026-09-27 and that…
The page documents multiple pricing and capability details for AI search APIs, including Exa pricing by tier (instant $4/1,000 requests; fast/auto $7; Deep Search $12–$15; up to 10 results with page contents; $1/1,000 for each result past 10), Brave Search API pricing ($5 per 1,000 requests for Web Search and LLM…
(Photon is Perplexity’s Rust retrieval and ranking engine; replaces open-source engine; production traffic uses Photon; Fast Search mode available via POST /search with search_type: "fast"; latency p99 fell from ~800 ms to ~65 ms; hosted API pricing $1 per 1,000 requests; Photon not open source)
Parallel’s official Search API page now documents the selected fast-search mode as “fast” rather than “turbo” in the previous snapshot. Because search modes are part of the API’s endpoint behavior and system record, this is a material mode-naming or configuration change requiring verification in the provider…
Parallel materially changed its Search API offering: the fast search mode is now labeled “turbo” instead of “basic,” and the displayed cost decreased from $5 to $1 per 1,000 requests. The page continues to describe multiple search modes with different latency/depth tradeoffs, making this a material pricing and…
Openbenchmarks materially updated its independent web-search evaluation board on September 15. The update added Nimble lite and standard configurations across company-news, coding-agent, and multi-turn search benchmarks, plus You.com highlights with and without knowledge=core. The board now reports You highlights at…
Parallel materially updated its Search API offering with four search modes: Turbo at about 200 ms and $1 per 1,000 requests, Fast at about 700 ms and $1 per 1,000, Basic at about 1 second and $5 per 1,000, and Advanced at about 3 seconds and $5 per 1,000. The page also states a 600-requests-per-minute limit and…
Perplexity published a major Advanced Deep Research update. The system now searches more sources, cross-references information, processes uploaded documents, runs calculations through an improved code sandbox, and browses harder-to-reach web sources. It also adds clarifying questions, follow-up questions during…
OpenAI published a Perplexity deployment update stating that Perplexity uses GPT-6 Astra to write communications, modify software, test applications, and monitor production systems. Perplexity also reports that Astra improves programs that search the web and internal information, and that it can be trusted with…
DeepSearch announced a new AI research platform for professionals that combines public-web research, source-linked findings, and contextual follow-up questions in one workspace. The platform is positioned as a source-backed research system that lets users inspect original material, compare accounts, identify evidence…
Openbenchmarks added new quantitative headline results to its independent web-search benchmark: Exa fast leads factual lookup at 99.3% across 300 questions, Exa deep leads hard retrieval at 83.0% across 100 tickets, and Parallel basic leads multi-hop search at 46.5% F1 across 45 questions. The page also clarified…
Openbenchmarks materially updated its independent web-search benchmark. The September 11 update replaced Tavily Fast with Tavily Basic on the search-only hard-retrieval board, evaluated 100 tasks across three repeats, replaced Tavily ultra-fast with Basic and Advanced on the company-news benchmark, and changed…
OpenAI released the Agents API in public beta. The API provides a managed Codex harness for session orchestration, context compaction, recovery, durable multi-turn sessions, progress streaming, custom tools and MCP servers, with either OpenAI-hosted sandboxes or supported external infrastructure. This is a new…
Openbenchmarks published an independent web-search benchmark update covering factual lookup, hard retrieval, and multi-hop search. The page reports a September 9, 2026 re-evaluation of TinyFish after updates to its GA Fetch endpoint, and provides quantitative provider comparisons using accuracy, answer recall,…
Parallel materially expanded its official quality benchmarks with new direct comparisons against Perplexity, Exa Auto, and Tavily across SimpleQA Verified, FRAMES, BrowseComp, and WideSearch. The revised methodology states that every competitor was evaluated at both low-cost and frontier tiers using the same agent,…
Parallel’s official benchmarks page now reports a new September 9, 2026 evaluation round for its Search API and Task API. The update adds fresh quantitative results across SimpleQA Verified, FRAMES, BrowseComp, and WideSearch, comparing Parallel Fast, Turbo, Advanced, and Basic tiers against OpenAI Web Search.…
Fortune reported that OpenAI repeatedly changed GPT-6 Astra’s published evaluation results after launch. Astra’s reported hallucination rate changed from 4.2% to 2% and later returned to 4.2%; its ARC-AGI-3 score changed from 98.6% in a pre-publication draft to 99.99% in the live post; and GPT-5.6 Sol’s ExploitBench…
OpenAI introduced GPT-6 Astra in a limited organizational rollout as a new model for research, coding, computer use, and complex multi-step work. The release adds the GPT-6 Astra model to the Responses and Chat Completions APIs, while the Responses API gains material agent-workflow controls including asynchronous…
keenable.ai (read 2026-08-26): Agent Builder tier $4 / 1000 requests; Frontier tier $1 / 1000 requests at 100 RPS+; <250 ms p95 (US East); 100B+ documents.
Research guide (read 2026-08-26): research_effort frontier at $1,200 per 1k calls, background=true required (sync → 422), 30 s–12,000 s with p50 300 s. Other tiers: lite $12, standard $50, deep $100, exhaustive $450 per 1k. Not wired in the arena.
OpenAI's deprecations page (read 2026-08-26) has no deep-research entry. The 2026-12-11 shutdown announced 2026-06-11 covers o3-2025-04-16, o3-pro-2025-06-10 and gpt-5-*-2025-08-07 snapshots → gpt-5.6-sol / terra / luna. gpt-5.5-pro is live at $30 / $180 per 1M. research.site's earlier “o3-deep-research retires 2026-12-11” item was wrong and is withdrawn.
Changelog 2026-08-24: citations and confidence scores are now emitted per list element (separate FieldBasis entries) instead of once per field, without the beta header.
OpenAI changelog 2026-08-21: gpt-5.6-sol pricing updated to $4.00 / 1M input and $20.00 / 1M output.
Changelog 2026-08-21: Search API mode "fast" at $1 per 1,000 requests with ~700 ms average latency; Turbo mode (2026-07-13) is $1 per 1,000 with p50 250 ms; basic / advanced stay $5 per 1,000. Task API processor prices unchanged ($5 lite → $2,400 ultra8x per 1k runs).
Changelog 2026-07-21: Responses API for latency-sensitive use, 5–60 s response times, structured outputs and full citations; synchronous, alongside the async Task API.
One GPT-6 Astra agent, five web tool stacks. Exa and built-in web tie at 11/26 on LiveBrowseComp, with Keenable one answer behind. Explore accuracy, run duration, and every public task verdict.
Claude Code vs Codex CLI under three search backends (builtin, Exa MCP, Valyu MCP) on a frozen 30-task LiveBrowseComp subset. claude-exa wins at 60%; the search backend moves accuracy up to 33 points, the agent at most 7.
Five deep-research APIs plus a Perplexity calibration anchor on a frozen 30-task ResearcherBench subset. Coverage is a three-way race; faithfulness splits the field in two; parallel has the best overall profile.
Nine deep research APIs on a frozen 20-task subset of DeepResearch-Bench-II. Exa effort tiers against each other, then one flagship tier per provider.