Which LLM Should I Use?
Describe the job. The page shows what it read, what it assumed, and a shortlist you can change one answer at a time.
Describe the work, where it can run, your budget, and the input size. This selector recommends models; it cannot carry out the task, open links, run code, or access your files and accounts.
The reader only fills in the form below. It recognizes requirements and models you mention; it never sees a score or chooses the shortlist. The shortlist is computed on this page from public evidence, every answer is editable, and the link carries your answers.
The answers under the shortlist
Tap any chip. The shortlist reruns at once.
Kind of work
Open. Pick one.
Processing
Open. Pick one.
Budget
Open. Pick one.
Compare one stated workload: 1M input + 200k output tokens, before caching, retries, tools, and special tiers.
Input size
Open. Pick one.
What would make a model wrong
Open. Pick one.
Choosing
Assumed. Tap to change.
Describe the job above, or pick the kind of work, and the shortlist appears here.
How the LLM selector makes its recommendation
The selector covers 31 tasks for everyday users and builders. Each task uses a relevant evidence source, with its limitations stated. Budget, context, and local-deployment requirements narrow the candidates. Scores are never increased because a model is open weight or uses a particular reasoning style.
A description is read by a decision model that fills in the form and nothing more: it reads requirements and recognizes models you mention, while the shortlist comes from the dataset. Where it read low confidence, the job is flagged for confirmation and the runner-up jobs are one tap away. Answers you did not state are assumed defaults, marked as such, and the shortlist stays provisional while answers or required setup features need checking.
Voice, image generation, specialist research, and other unsupported tasks receive a clear evidence gap instead of a recommendation built from unrelated scores. Compare current API pricing, inspect visual and document evidence, or read the scoring methodology.
Once you have a shortlist, draft a task prompt or improve an existing prompt to use in your comparison. Running a model you may replace? The alternative finder shows the quality and price delta against it.
Method: bench-align-v5.8-2026-09-30.
Questions
How does the LLM selector work?
Describe the job, or pick it from the list. The page shows what it read, fills every remaining question with a stated default, and computes a shortlist at once from the public dataset: a relevant category estimate or observed task result, filtered for the stated budget, context, and deployment needs. Each answer you change reruns the shortlist and states what moved. There is no hidden weighted fit score.
What does the reader do with my description?
A decision model fills in the same multiple-choice form you see on the page, and nothing else. It recognizes requirements and models you mention; it never sees a score or chooses the shortlist. Every answer it fills is marked as coming from your description, and you can change any of them. The shortlist is computed on this page from the dataset, exactly as it would be from clicks.
Why is the shortlist marked provisional?
A shortlist is provisional while any answer under it is an assumed default rather than something you said or chose, or while the job was read with low confidence. The assumed answers are marked. Confirm or change them and the label goes away.
Which LLM should I use for coding?
Choose Build or fix software, then the kind of work you do. Coding-agent results depend on the model, editor or agent harness, and effort setting. Compare the suggested models on representative repository tasks before switching.
What is the best free LLM?
A free assistant plan, an open-weight model, and free hosting are different things. For local deployment, the selector limits candidates to open weights. For assistant plans, model access and message limits must be checked with the provider; API prices are not subscription prices.
Why might the selector return no recommendation?
A task may lack suitable benchmark evidence, or no measured candidate may satisfy your constraints. Missing prices cannot pass a price limit. The selector keeps these gaps visible instead of silently relaxing your requirements.
Can I ask why a particular model is missing?
Name it in the description. The page then states where that model stands for the job: on the shortlist, filtered out by a constraint you stated, below the cut with its score, or without evidence for this job.
Can I share or reuse a result?
Every answer is written into the page address, so the link reopens the same shortlist against the current dataset. The same answers work as query parameters on the /api/tools/recommend endpoint for scripts and agents.