Answer: SWE-grep models power Windsurf Fast Context and come in two variants: SWE-grep, a high-intelligence variant optimized for complex retrieval tasks, and SWE-grep-mini, an ultra-fast variant serving at over 2,800 tokens per second. Both are trained using reinforcement learning for parallel tool calling and efficient codebase navigation. Status: Check failed Entry: tool/windsurf/models/swe-grep-models Revision: 1 Digest: sha256:7306e8f323e0207b (changes whenever the content changes) Timeline: Created: 2026-11-13T03:58:39Z Last changed: 2026-11-13T03:58:39Z Last verified: 2026-11-13T03:58:39Z Last attempt: 2026-11-19T13:37:46Z (failed) Next check: 2026-11-19T14:37:46Z Check interval: Daily Expected change: No fixed schedule Data: Field | Value | Unit | Source | As of model family | SWE-grep | | [1] | 2026-11-13 high-intelligence variant | SWE-grep | | [1] | 2026-11-13 ultra-fast variant | SWE-grep-mini | | [1] | 2026-11-13 mini throughput | 2,800 tokens per second | tokens per second | [1] | 2026-11-13 parallel tool calls per turn | 8 | tool calls | [1] | 2026-11-13 maximum turns | 4 | turns | [1] | 2026-11-13 training method | reinforcement learning | | [1] | 2026-11-13 tools | grep, read, glob | | [1] | 2026-11-13 Sources: [1] Windsurf. "Fast Context". https://docs.devin.ai/desktop/context-awareness/fast-context Role: primary. Accessed: 2026-11-13T03:58:39Z Quote: "Fast Context is powered by the SWE-grep model family" (Supports: model family) Quote: "SWE-grep : High-intelligence variant optimized for complex retrieval tasks" (Supports: high-intelligence variant) Quote: "SWE-grep-mini : Ultra-fast variant serving at over 2,800 tokens per second" (Supports: ultra-fast variant) Quote: "serving at over 2,800 tokens per second" (Supports: mini throughput) Quote: "They execute up to 8 parallel tool calls per turn over a maximum of 4 turns" (Supports: parallel tool calls per turn) Quote: "over a maximum of 4 turns" (Supports: maximum turns) Quote: "Both models are trained using reinforcement learning to excel at parallel tool calling and efficient codebase navigation." (Supports: training method) Quote: "The models use a restricted set of cross-platform compatible tools (grep, read, glob)" (Supports: tools) Published: 2026-11-13T01:50:40Z Snapshot SHA-256: 21a10787f2dbd1d77f9edd740826c2a8e7a8ccb25d81180d91605d6fd188246c (captured 2026-11-13T03:58:37Z) Verification: Method: Automated extraction from [1] with deepseek-flash (JSON output, thinking disabled). Every value was checked by code to appear verbatim in a quoted passage of [1]. Cross-check: not-available at 2026-11-13T03:58:39Z. No independent second source is configured for this entry Last successful check: 2026-11-13T03:58:39Z Last attempt: 2026-11-19T13:37:46Z (failed) Notes and limitations: Extracted automatically from a single official source (section "​ SWE-grep Models"). Human-readable page: https://docs.windsurf.com/context-awareness/fast-context.md Related: https://latestn.com/tool/windsurf/permissions-and-security/configure-sso-authentication How do I configure SSO authentication for my Windsurf team? https://latestn.com/tool/windsurf/pricing-and-limits/adaptive-pricing How is Windsurf Adaptive priced? https://latestn.com/tool/windsurf/pricing-and-limits/automatic-credit-refills How do automatic credit refills work in Windsurf? https://latestn.com/tool/windsurf/models/how-adaptive-works How does Windsurf Adaptive choose which AI model to use? https://latestn.com/tool/windsurf/models/quick-review-models Which AI models are available for Windsurf Quick Review? https://latestn.com/tool/windsurf/models/select-adaptive How do I select Adaptive in Windsurf? https://latestn.com/tool/ All entries in AI tools Change log: Time | Event | Change | Source 2026-11-13T13:31:48Z | check-failed | Extraction did not pass verification: section "​ SWE-grep Models" not found in the source (the page may have been reorganized) | — 2026-11-13T03:58:39Z | added | Entry created | — Changes for this entry as a feed: https://latestn.com/tool/windsurf/models/swe-grep-models/latest Challenge this entry: Open challenges: None How: If a value here is wrong or out of date, send what is wrong and a link to evidence. Challenges are queued and checked against the evidence; they do not change the entry by themselves. If the entry changes as a result, the change log says so. Form: https://latestn.com/dispute/tool/windsurf/models/swe-grep-models POST: https://latestn.com/dispute/tool/windsurf/models/swe-grep-models with claim, evidence_url, and optionally field and contact (form-encoded or JSON) GET: https://latestn.com/dispute/tool/windsurf/models/swe-grep-models?claim={what is wrong}&evidence_url={link} How to cite: LatestN. (2026). What are SWE-grep models in Windsurf Fast Context? Revision 1, last verified 2026-11-13T03:58:39Z. https://latestn.com/tool/windsurf/models/swe-grep-models (accessed 2026-11-21T09:59:15Z). @misc{LatestN_tool_windsurf_models_swe_grep_models_r1, title = {What are SWE-grep models in Windsurf Fast Context?}, author = {{LatestN}}, year = {2026}, howpublished = {\url{https://latestn.com/tool/windsurf/models/swe-grep-models}}, note = {Revision 1; digest sha256:7306e8f323e0207b; last verified 2026-11-13T03:58:39Z; accessed 2026-11-21T09:59:15Z} } Reuse: License: CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/) Attribution: Copying and adapting are allowed, including commercially, if you credit LatestN and link to the entry you used. Each entry ends with a ready-made citation. URL: https://latestn.com/tool/windsurf/models/swe-grep-models