Use case

Long-context analysis

Working with prompts and corpora that exceed the small-context tier (≥200k tokens). The relevant verified fields are context window, max output tokens, and the pricing reference rows that apply at the prompt-size tier the vendor publishes.

Verified fields used

context window
max output tokens
pricing references (incl. >200k tier where the vendor publishes one)
lifecycle status
source citations

Long-context workloads — codebases, multi-file transcriptions, multi-document research — depend on the vendor's published context window. The verified context-window field on each model record is backed by the vendor's own model page, and it is the only context number this site asserts.

Three caveats every reader should hold in mind:

A large window is not a quality claim. It means the model can accept a long prompt; it does not mean every token is equally well used. Comprehension degrades with prompt size on every model published today.
Long context can cost more. Vendors that publish a prompt-size pricing tier (Google's >200k surcharge is the live example) charge differently above a threshold. Cross-check the relevant pricing tier on /pricing and remember that pricing is a source-backed reference, not a live quote.
Max output limits are usually much smaller. A million-token context window does not imply a million-token output. Each model records max output separately; unverified output limits remain null.

Open a candidate's detail page from the shortlist below and inspect the verified context-window, max-output, and pricing reference rows side-by-side with the vendor citation. For a paired side-by-side reference between two candidates, open /compare after selecting two finalists.

Shortlist

Shortlist for long-context analysis (9)

Generated from the typed local data layer. Shortlist order: verified field count → active lifecycle → source count → name. Not a ranking.

Open in selection workspace →

Open comparison builder for this use case → Pre-seeded with the top 4 shortlist candidates; these are candidates, not picks.

Create evidence brief for this use case → Markdown or JSON export of the same selection — evidence, not a recommendation.

Model	Provider	Lifecycle	Context	Modalities	Sources	Freshness	Next action
Gemini 2.5 ProVerified	google	active	1,048,576	text-in, image-in, audio-in, video-in, text-out	3	Fresh	Review the verified pricing reference and re-check the freshness state before projecting cost.
Claude Haiku 4.5Verified	anthropic	active	200,000	text-in, image-in, text-out	2	Fresh	Review the verified pricing reference and re-check the freshness state before projecting cost.
Claude Opus 4.7Verified	anthropic	active	1,000,000	text-in, image-in, text-out	2	Fresh	Review the verified pricing reference and re-check the freshness state before projecting cost.
Claude Sonnet 4.6Verified	anthropic	active	1,000,000	text-in, image-in, text-out	2	Fresh	Review the verified pricing reference and re-check the freshness state before projecting cost.
Claude Opus 4Verified	anthropic	deprecated	200,000	text-in, text-out	2	Fresh	Historical record — open the model page to read the lifecycle note.
Llama 4 MaverickVerified	meta	active	1,000,000	text-in, image-in, text-out	1	Fresh	Open the model page to inspect verified fields and source citations.
Llama 4 ScoutVerified	meta	active	10,000,000	text-in, image-in, text-out	1	Fresh	Inspect the hosted-availability record and the third-party hosted pricing reference on the provider page.
Mistral Large 3Verified	mistral	active	256,000	—	4	Fresh	Review the verified pricing reference and re-check the freshness state before projecting cost.
DeepSeek V4 ProVerified	deepseek	active	1,000,000	—	3	Fresh	Review the verified pricing reference and re-check the freshness state before projecting cost.