AI Developer Tools

What it helps you do

Develop formal mathematical proofs and explore verification of software or hardware specifications with an AI reasoning agent.

Account access is required and designated usage consumes prepaid credits. Check account pricing, credit expiry and eligibility before starting. Outputs require independent review against the supplied specification. Terms allow model training on submitted data unless the account opts out for future submissions; do not submit personal or prohibited data.

What it helps you do

Open-source memory and tracing layer that captures coding-agent sessions across tools and turns reviewed workflows and corrections into reusable skills.

Local open-source use and a hosted Cloud plan are free; Cloud includes 50 findings. Pro is listed at $20/month billed annually, with enterprise pricing by quote. Interactive setup preselects Cloud; choose Local to keep captured sessions on your machine. Full Cloud memory and skills access varies by plan. Cloud research and model training are enabled by default and can be disabled.

What it helps you do

Run browser agents and managed browser sessions for website workflows through a hosted dashboard and developer API.

Cloud account and API access use prepaid pay-as-you-go credits. Eligible new social signups receive $1 once; minimum top-up is $5. Agents cost model usage plus a 20% service fee, browser time and traffic. Browsers are $0.02/hour; default residential proxy traffic is $5/GB. SUPERFAST is separately waitlisted.

What it helps you do

Provide cloud browsers and session visibility for AI agents, browser automation and web applications.

Developer dashboard and API; integration required. Free plan lists 1 browser hour/month, 3 concurrent sessions, a 15-minute session limit and limited agent/search/fetch allowances. Developer plan is $20/month with 100 browser hours and usage overages. Proxy traffic and model usage have separate allowances and charges; paid plans are not unlimited.

What it helps you do

Coding editor and agent platform for writing, reviewing, and coordinating code changes locally or in the cloud.

A no-card Hobby tier has limited agent requests. Paid plans add usage and cloud/team features; extra model usage and Bugbot billing have separate limits. Enterprise includes an AI code-tracking API.

What it helps you do

Give AI agents cloud sandboxes for running code, using software and working in repeatable computing environments.

Developer account, console and SDK integration. Hobby has no base subscription but usage charges apply, with $100 in one-time trial credit and no card required to start. Hobby sessions are limited to 1 hour and 20 concurrent sandboxes. Pro is $150/month plus usage with higher limits. Runtime bills CPU and memory by time; model-provider costs are separate.

What it helps you do

Search API for agents that returns live web results, page text, and highlights, and can run longer research jobs.

Free includes $10 in credits every month plus a $10 onboarding bonus, with no payment method. Search is listed from $4, $7, and $12 per 1,000 requests depending on the mode. Contents are $1 per 1,000 pages. Agent runs are listed from $0.012 to $1.00. Monitors are $15 per 1,000 requests. Exa 3.0 is dated 19 October 2026 on Exa's blog.

What it helps you do

Search the web, turn pages into Markdown or structured data, and add browser interaction to AI applications.

Hosted dashboard and developer API. Free plan lists 1,000 credits/month without a card; paid Hobby is $19/month or $16/month billed annually. Action and format choices affect credits. Agent API is a preview with 5 free daily runs and dynamic pricing. Paid auto top-ups and specialist data providers can add charges.

What it helps you do

Coding assistant for code suggestions, codebase questions, changes, reviews, and assigned development tasks.

Free and paid plans across supported IDE, GitHub, and terminal workflows; features and usage depend on plan and organization policies.

Other offerings from this vendor: Microsoft Copilot.

What it helps you do

Graphical workspace for fine-tuning language models, comparing experiments and evaluating model outputs.

Apache-2.0 software; requires Ubuntu and a recent NVIDIA GPU. Larger models typically need 24 GB GPU memory; infrastructure and model licenses are separate.

What it helps you do

Runs an app in a sandbox, looks for bugs and vulnerabilities, and reports only issues it can reproduce with a failing test.

A free trial does not require a credit card, and open-source projects can request free credits. Paid prices are not published. The site also offers a demo. The public launch date recorded here is 22 October 2026.

What it helps you do

Discover and share AI models, datasets, and apps, with hosted inference and paid compute options.

Free Hub collaboration and basic Spaces hosting; dedicated hardware and inference are separately billed. Each model and dataset has its own license.

What it helps you do

Work with persistent agents that remember context, use tools, and run coding or research tasks locally or in the cloud.

Web account or Letta Code CLI; CLI docs require Node.js 22.19+. Free personal plan supports 3 stateful agents with limited model usage and external keys or coding plans. Pro is $20/month with quotas and paid overage. API plan is separate: $20/month plus agent, tool-execution and model usage. External providers can charge separately.

What it helps you do

Turns production agent traces into smaller models you can evaluate, serve through an OpenAI-compatible API, or run yourself.

Free is $0 and includes a one-time $50 credit and one seat per project. Pro is $120 a month, with $120 in credits each month and unlimited seats. Enterprise is custom. One credit is $0.01. The public platform launch date recorded here is 27 October 2026.

What it helps you do

Give AI applications web search, content extraction, monitoring and research through developer APIs.

Developer account and API key. Pricing advertises up to $80 in signup credits. Eligible organizations with a card receive $5 monthly credits; unused credit expires, with one organization per card and marketplace/postpaid exclusions. Additional usage is billed by API and selected processing tier. Your application’s own model and other services can cost extra.

What it helps you do

Mobile automation service that lets AI agents operate provisioned Android phones through MCP, an API or a CLI, with screen observations and recorded action outcomes.

Paid from first use; no self-service free tier. Cloud Android starts at $20/month or $0.024/minute while rented. The website lists real iPhones at $99/month by reservation; current device documentation says robot-hand iPhones are call-only and outside API, MCP and CLI access. Copilot sends phone-screen content to AI providers; deleting chats does not erase stored records.

What it helps you do

Build coding and reasoning applications with the Beam model through an API, agent tooling, or a self-hosted deployment.

Public-beta API access requires an account and API key. Check current limits and billing in the platform; a complete public price schedule was not verified. Self-hosting the published weights requires suitable compute and separate operating costs.

What it helps you do

Hosted AI models and deployment tools for generating images, video, audio, and text through APIs or web playgrounds.

Replicate is a distinct Cloudflare product with usage-billed hosted model runs. Most private deployments also bill setup and idle time. Custom models require technical setup; model licenses and rates vary.

What it helps you do

Shared pages where agents publish docs, dashboards, and mockups that people can comment on, version, and sign.

Free is $0, needs no card, and includes 200 files and 25 MB per deploy plus 7 million agent tokens a month. Pro is $4.50 a month, Max is $9 a month, and Team is $10 per member a month. Enterprise is contact. The public launch date recorded here is 27 October 2026.

What it helps you do

Runs synthetic or real users through a product, including from coding tools, and summarizes what they struggle with.

The free plan is $0 a month, with 5 AI user tests, 50 AI surveys, and one workspace, and it does not require a card. The pricing page does not list dollar prices for other tiers. Enterprise is contact sales. The homepage says synthetic tests start at $1. TheySaid 4.0 is the release dated 27 October 2026 for this listing.

What it helps you do

Fine-tune language models with supervised learning or reinforcement learning while a managed service runs the training infrastructure.

An account and developer integration are required. Training and sampling are billed by token usage, with separate checkpoint-storage charges. Inkling serverless inference is beta; intensive production access has a separate waitlist. Check current models and rates before starting a run.

What it helps you do

Desktop terminal and coding-agent workspace for developing software and coordinating command-line tasks.

Free terminal with limited cloud features; agent usage uses paid credits or your own provider. The client is open source; cloud Factories and enterprise APIs have separate access terms.

What it helps you do

Track model experiments and artifacts, and evaluate or debug AI applications with traces and evaluations.

Part of CoreWeave Forge. Free covers personal development within storage/ingestion limits; Pro offers a 30-day trial with team eligibility rules. Inference and other services have separate usage charges.

What it helps you do

Add web search, page extraction, cited answers, and multi-step research to an application or AI agent through hosted APIs and MCP access.

API use requires platform account setup and an API key, with metered charges by endpoint or returned page. The pricing page documents limited free MCP search without an API key and introductory API credit; this does not make every API endpoint free. Confirm credit and billing conditions in the platform.