Grounding with Bing Search vs Perplexity Search API

Grounding with Bing Search 61

Read as a server-side search tool inside an agent run. The tools return a model's answer with citations rather than raw results, and search_context_size sets how much context web search uses. A caller can't size or inspect the result set itself, since the Bing tools never return their output. Custom Search confines grounding to chosen domains without code (18 of 25). count, freshness, market and language filters, and domain restriction through Custom Search. No pagination or offset (13 of 20). Troubleshooting tables give causes and fixes for connection ID, RBAC and 429 problems, but the tools document no error codes (8 of 20). Searches only read, so a retry is safe apart from the per-transaction charge. No MCP annotations are documented for the toolbox's web search tool (10 of 20). The web_search route needs only a project, a model deployment and the Foundry User role. bing_grounding also needs a paid Bing resource and a project connection. SDKs in Python, C#, JavaScript and Java (12 of 15). Prompt caching gets no hits with the Bing tools, because they add a timestamp to every request. 18 + 13 + 8 + 10 + 12 = 61.

Perplexity Search API 85

max_results 1 to 20, max_tokens and max_tokens_per_page cap the extracted text, and results are title, url, snippet and two dates. The MCP server has four tools (23 of 25). Domain, language, recency, publication and update date filters and country, with up to five queries in one request. No pagination or offset (16 of 20). 429 bodies name the limit hit, the SDKs map status codes to typed errors, and 402 marks an empty balance. Search-specific error bodies beyond 422 validation are not documented (13 of 20). Searches only read, failed and rate-limited requests are not billed, and the MCP server sets readOnlyHint true, destructiveHint false and idempotentHint false on every tool (18 of 20). Only query is required. Official Python and TypeScript SDKs, a CLI and a Python Search SDK (15). 23 + 16 + 13 + 18 + 15 = 85.

Grounding with Bing Search 79

Callers authenticate to the Foundry project with Microsoft Entra ID tokens and RBAC roles (Foundry User to run agents, Foundry Project Manager to add connections), with just-in-time Contributor through Entra PIM for provisioning. The Bing resource key sits in a project connection and the agent never holds it. No key in a URL (28 of 30). The tools only read. Custom Search confines grounding to configured public domains, and an admin can turn web search off for a whole subscription. There is no per-agent read-only toggle to set, since the tools have no write path (15 of 20). The web search page tells developers to treat web search results as untrusted input and to keep secrets and personal data out of prompts, and the tool best practices page says the same of all tool outputs. Raw results are not passed back to the caller (10 of 15). Run steps show each tool call and the Bing query sent, and Azure's activity log covers resource changes. No per-query audit log for Bing was found (10 of 15). Microsoft runs a coordinated vulnerability disclosure policy and bounty programmes through MSRC. microsoft.com's security.txt passed its Expires date on 23 September 2026 and still carries it on 6 November, and the Azure compliance attestations are not shown to cover a service whose data leaves the compliance boundary (16 of 20). 28 + 15 + 10 + 10 + 16 = 79.

Perplexity Search API 56

Project API keys, shown once, revocable, and generated and revoked through two documented endpoints with a rotation guide. No per-key scopes or spending limits were found. The hosted MCP server also takes Sign in with Perplexity (OAuth). Keys go in the Authorization header only (22 of 30, as on the Decisions API listing). Search only reads, and MCP clients can disable tools, but any key reaches every API on the account, including the Agent API's sandbox and MCP tools (8 of 20). Snippets are third-party page text, and no guidance on treating them as untrusted was found in the docs, the MCP server or its SECURITY.md (0 of 15). Responses carry a request id, and the console shows usage by key. The Audit API covers enterprise organisations' web app events with its own keys, not API calls (8 of 15). security.txt valid until 31 December 2027 linking a security page, a SOC 2 Type II report, a 2025 HIPAA gap assessment and a CAIQlite listed on the trust centre, and a SECURITY.md on the MCP server that takes private reports through GitHub advisories or security@perplexity.ai. The trust centre is drawn by script and wasn't read (18 of 20). 22 + 8 + 0 + 8 + 18 = 56.

Grounding with Bing Search 68

The tools are typed in the official SDKs (BingGroundingTool, BingCustomSearchPreviewTool, WebSearchTool) in Python, C#, JavaScript and Java, and the toolbox exposes web search as an MCP tool. No public OpenAPI description of the Responses tool types was confirmed from the pages read (10 of 25). Microsoft Learn serves each page as Markdown with accept=text/markdown, though learn.microsoft.com/llms.txt answers 404 (10). The pages say what each of four web grounding routes is for and when to choose it, when to use Custom Search instead, and not to ask the model to summarise a whole page (17 of 20). Parameters are typed with defaults and limits in prose, count 1 to 50 with default 5, freshness values and a date-range form, market and language formats, search_context_size low, medium or high (11 of 15). Examples in Python, C#, TypeScript, Java and REST with expected output, and troubleshooting tables for connection ID, authentication and 429 problems, but no error codes or bodies for the tools (10 of 15). Requests carry api-version=v1 on the toolbox, SDK changelogs are dated and a monthly What's New page lists new articles, the latest for August 2026 (10 of 15). 10 + 10 + 17 + 11 + 10 + 10 = 68.

Perplexity Search API 89

OpenAPI 3.1.0 at docs.perplexity.ai/openapi.json describes POST /search with typed request and response schemas, listed in llms.txt (25). llms.txt indexes every page, each with a Markdown twin (10). The overview says what a result holds, that snippets can be empty, when to use the Search API instead of the Agent API, and that it takes queries not URLs and does not re-fetch pages (19 of 20). search_type is an enum, max_results, max_tokens and max_tokens_per_page carry minimum and maximum, country and language codes carry lengths, and query is a string or an array. The recency filter is an enum in the docs, and date filters are MM/DD/YYYY strings (14 of 15). Python, TypeScript and cURL examples with a recorded response. The spec documents only 200 and 422 for /search, and status codes are listed on the SDK error-handling page and the rate-limit page rather than for Search (11 of 15). A public changelog with several entries a month, dated by month only, and an unversioned /search path (10 of 15). 25 + 10 + 19 + 14 + 11 + 10 = 89.

Grounding with Bing Search 68

Graded on the hosted lines, for the Foundry Agent Service tools agents call. Azure status page with post-incident reviews and a Foundry Agent Service component (20). The history page read on 6 November lists one review in the 90 days from 8 August that names Foundry Agent Service, intermittent failures and latency in Sweden Central from 10:03 to 15:58 UTC on 29 September 2026. It was one region and partial, and the review doesn't name Grounding with Bing, so minor only, the reading the Azure Translator listing got for the same review (20). The pricing page publishes 150 transactions a second and 1 million a day for each Grounding with Bing resource (15). The web search troubleshooting table says to answer 429 with exponential backoff and retries. No Retry-After or tool-specific overload guidance was found, and searches have no side effects (8 of 15). No SLA naming Grounding with Bing was found. The terms make it a First Party Consumption Service apart from the Online Services terms, and the Online Services SLA document wasn't opened (0). Grounding with Bing Custom Search is marked preview and Grounding with Bing Search is not, but no page read states general availability for it or for the web_search tool, and the .NET SDK is in preview (5 of 10). 20 + 20 + 15 + 8 + 0 + 5 = 68.

Perplexity Search API 80

Graded on the hosted lines, for POST /search and the hosted MCP server. status.perplexity.com, on incident.io since 10 September 2026, has an API component with an RSS feed and a JSON summary. There is no separate Search component (20). In the 90 days from 8 August the feed lists 'Investigating API issue' on 13 August, resolved with no detail, a usage-based billing issue on 3 September, and on 9 October elevated errors on the website, app and Computer, where the API 'remained operational throughout'. Minor only, with the caveat that history before the migration may be incomplete (20). The Search API is limited to 50 query units a second with a burst of 50 on every tier, and the other APIs publish QPS and token limits per tier (15). 429 carries Retry-After and an error.type, the SDKs retry 429 and 5xx twice honouring Retry-After, the Search best-practices page shows backoff, and failed or rate-limited searches are not billed. Searches have no side effects (15). The FAQ says Perplexity does not guarantee uptime, failure frequency or recovery time, and the terms carry no SLA (0). No beta label on the Search API, launched on 25 September 2025 (10). 20 + 20 + 15 + 15 + 0 + 10 = 80.

Grounding with Bing Search 20

No x402, MPP or L402 (0). $14 per 1,000 transactions for Grounding with Bing Search and Custom Search, published on microsoft.com without a login (20). No free tier for Grounding with Bing, and an Azure subscription needs a card, as on the Azure Translator listing (0). A person signs up for Azure, creates a Foundry project and, for bing_grounding, a Bing resource (0).

Perplexity Search API 25

No x402, MPP or L402 in the docs, pricing page, OpenAPI or llms.txt (0). $5 per 1,000 successful requests, $1 with Fast Search, published without a login, with failed and rate-limited requests not billed (20). Usage is prepaid with a card on the billing page, and no free allowance was found. The console playground needs no key but is a page for a person, not an API allowance (0). A person signs up in the console. The Stripe Projects route provisions a project, key and credits from a terminal, but needs a person's Stripe account with a payment method and a $10 minimum purchase (5 of 20, as on the Decisions API listing). 0 + 20 + 0 + 5 = 25.