---
title: "Models and context"
canonical: "https://canmyagentuse.com/features/models-and-context"
contentKind: "feature"
locale: "en"
description: "Compare model choice, reasoning controls, context budgets, compaction, fallback, and prompt caching as separate harness capabilities."
llmSummary: "Models and context is a capability family covering model choice, reasoning controls, usable token budgets, context visibility, compaction, fallback, local inference, and prompt-cache behavior."
publishedAt: "2026-08-28T00:00:00.000Z"
updatedAt: "2026-08-28T00:00:00.000Z"
verifiedAt: null
tags: ["models","context","tokens","caching"]
---

# Models and context

Models and context is a capability family covering model choice, reasoning controls, usable token budgets, context visibility, compaction, fallback, local inference, and prompt-cache behavior.

- HTML: https://canmyagentuse.com/features/models-and-context
- JSON: https://canmyagentuse.com/api/v1/features/models-and-context.json
- Markdown: https://canmyagentuse.com/features/models-and-context.md

Terminology basis: Catalog grouping. This organizes related compatibility questions and is not itself a support claim.

A harness is not a transparent pass-through to a model. It can select or route models, reserve tokens for hidden instructions and tools, enforce a lower input cap, compact earlier turns, retry on another model, and cache only certain prompt segments.

This family therefore avoids copying a provider model-card number into every product surface. Each child row asks a narrower question and keeps model, mode, plan, workspace policy, file-processing cost, tool-schema overhead, and observation date attached to the answer.

For context limits, useful records report the advertised maximum and the effective usable budget when those differ. For caching, they distinguish reuse itself from explicit controls and operator-visible telemetry.

## Atomic capabilities

- [Automatic context compaction](/features/automatic-context-compaction.md) — Continue long sessions by automatically summarizing or pruning older context.
- [Context compaction controls](/features/context-compaction-controls.md) — Trigger, inspect, configure, or disable context compaction.
- [Context usage](/features/context-usage-visibility.md) — Inspect consumed or remaining context budget during a conversation or run.
- [Context window](/features/context-window.md) — Record the documented context-window limit by product, model, and mode.
- [Local models](/features/local-models.md) — Run against a model hosted on the operator's machine.
- [Model fallback](/features/model-fallback.md) — Switch to another model after a documented failure condition.
- [Model selection](/features/model-selection.md) — Choose the model or documented model class used for a run.
- [Output token limit](/features/output-token-limit.md) — Record the maximum generated response budget by exact model and harness mode.
- [Prompt cache controls](/features/prompt-cache-controls.md) — Mark, retain, refresh, or invalidate cached prompt context.
- [Prompt cache telemetry](/features/prompt-cache-telemetry.md) — Inspect cache hits, misses, token counts, latency, or billing effects.
- [Prompt caching](/features/prompt-cache-reuse.md) — Reuse eligible repeated prompt content under documented caching behavior.
- [Reasoning effort controls](/features/reasoning-effort-controls.md) — Choose a documented reasoning depth, effort, or budget for a run.

## Current family progress by product

Family support is derived from atomic child records. No umbrella compatibility value is authored.

- [ChatGPT](/harnesses/chatgpt-web.md): 2/12 supported or partial; 2/12 reviewed; yes 2, partial 0, no 0, unknown 10
- [Claude](/harnesses/claude-web.md): 2/12 supported or partial; 2/12 reviewed; yes 2, partial 0, no 0, unknown 10
- [Gemini](/harnesses/gemini-web.md): 1/12 supported or partial; 1/12 reviewed; yes 1, partial 0, no 0, unknown 11
- [Copilot](/harnesses/copilot-web.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Grok](/harnesses/grok-web.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Grok Bot](/harnesses/grok-bot-desktop.md): 1/12 supported or partial; 2/12 reviewed; yes 1, partial 0, no 1, unknown 10
- [Perplexity](/harnesses/perplexity-web.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Le Chat](/harnesses/le-chat.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Devin](/harnesses/devin-web.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Replit Agent](/harnesses/replit-agent.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [ChatGPT](/harnesses/chatgpt-desktop.md): 4/12 supported or partial; 4/12 reviewed; yes 3, partial 1, no 0, unknown 8
- [Claude](/harnesses/claude-desktop.md): 2/12 supported or partial; 2/12 reviewed; yes 2, partial 0, no 0, unknown 10
- [Cursor](/harnesses/cursor.md): 3/12 supported or partial; 3/12 reviewed; yes 3, partial 0, no 0, unknown 9
- [OpenWork Desktop](/harnesses/openwork-desktop.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Copilot Chat](/harnesses/vscode-copilot.md): 5/12 supported or partial; 5/12 reviewed; yes 5, partial 0, no 0, unknown 7
- [Chrome WebMCP origin trial](/harnesses/chrome-webmcp-preview.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Windsurf](/harnesses/windsurf.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Zed Agent](/harnesses/zed-agent.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Continue](/harnesses/continue.md): 1/12 supported or partial; 1/12 reviewed; yes 1, partial 0, no 0, unknown 11
- [Cline](/harnesses/cline.md): 1/12 supported or partial; 1/12 reviewed; yes 1, partial 0, no 0, unknown 11
- [JetBrains AI](/harnesses/jetbrains-ai.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Warp](/harnesses/warp.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Claude CLI](/harnesses/claude-cli.md): 9/12 supported or partial; 9/12 reviewed; yes 7, partial 2, no 0, unknown 3
- [ChatGPT CLI](/harnesses/chatgpt-cli.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Codex CLI](/harnesses/codex-cli.md): 3/12 supported or partial; 3/12 reviewed; yes 3, partial 0, no 0, unknown 9
- [OpenCode](/harnesses/opencode.md): 1/12 supported or partial; 1/12 reviewed; yes 1, partial 0, no 0, unknown 11
- [Gemini CLI](/harnesses/gemini-cli.md): 6/12 supported or partial; 6/12 reviewed; yes 4, partial 2, no 0, unknown 6
- [Aider](/harnesses/aider.md): 1/12 supported or partial; 1/12 reviewed; yes 1, partial 0, no 0, unknown 11
- [Goose](/harnesses/goose.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12
- [Copilot CLI](/harnesses/copilot-cli.md): 4/12 supported or partial; 4/12 reviewed; yes 4, partial 0, no 0, unknown 8
- [Amp](/harnesses/amp-cli.md): 0/12 supported or partial; 0/12 reviewed; yes 0, partial 0, no 0, unknown 12