Skip to main content
AI Choice Engine

Cognition

SWE-grep

SWE-grep is a Cognition model with Not verified context, Not verified in / Not verified out per million tokens, last verified October 2, 2026. Open weights: not verified. Public-board scores are linked to the publisher; vendor-only figures stay in the watch-out, not on the board.

A specialist context-retrieval model used by Fast Context to identify relevant repository files and line ranges.

Recorded profile verification October 2, 2026Individual provider fields may changeAccess: restricted — Current docs identify SWE-grep as powering context retrieval and Fast Context.

AI model specification details
SpecificationSWE-grep
ProviderCognition
TierBalanced
Context windowNot verifiedUnverified
Max outputNot verifiedUnverified
Input / 1M tokensNot verifiedUnverified
Output / 1M tokensNot verifiedUnverified
WeightsNot verifiedUnverified
ParametersNot verifiedUnverified
Reasoning levelsNot verifiedUnverified
Modalitiestext
ReleasedOctober 16, 2025

Pricing tiers: In-product model access and billing vary by plan/account. Exact current model-specific rates were not recovered; the authenticated List Models response is the account source of truth. Historical free/speed offers are not current generic API prices.

Best for

  • Understanding the retrieval stage of the integrated coding workflow

Watch out

Exact context/output caps, stable generic inference API ID, downloadable weights/licence and model-specific numeric parameter count are unverified. The authenticated List Models endpoint returns account model-menu UIDs and plan-specific pricing; its placeholder examples are not live prices or general inference endpoints. Do not inherit base-model, harness or sibling benchmark scores; vendor speed/evaluation claims are not independent cross-model results. The launch's retrieval evaluations use the publisher's internal labelled repository dataset and custom tool/harness setup; they do not establish general coding quality.

Source receipts

Catalog figures for this model were checked against the following sources.

Common questions

SWE-grep

Answered from the verified figures on this page rather than general guidance.

What is SWE-grep best for?
SWE-grep is a balanced tier from Cognition. It suits understanding the retrieval stage of the integrated coding workflow. Exact context/output caps, stable generic inference API ID, downloadable weights/licence and model-specific numeric parameter count are unverified. The authenticated List Models endpoint returns account model-menu UIDs and plan-specific pricing; its placeholder examples are not live prices or general inference endpoints. Do not inherit base-model, harness or sibling benchmark scores; vendor speed/evaluation claims are not independent cross-model results. The launch's retrieval evaluations use the publisher's internal labelled repository dataset and custom tool/harness setup; they do not establish general coding quality.
Model comparison catalogBenchmark library