You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Model quality evolves quickly, and we tune the CLI defaults as the ecosystem shifts. Use this guide as a snapshot of how the major options compare today, and expect to revisit it as we publish updates. This guide was last updated on Wednesday, October 23rd 2025.
7
+
Model quality evolves quickly, and we tune the CLI defaults as the ecosystem shifts. Use this guide as a snapshot of how the major options compare today, and expect to revisit it as we publish updates. This guide was last updated on Thursday, December 4th 2025.
| 1 |**Claude Sonnet 4.5**| Recommended daily driver. Excellent balance of quality, speed, and cost for most development tasks. Current CLI default. |
16
-
| 2 |**GPT-5 Codex**| Fast iteration loops with strong coding performance. Great for implementation-heavy work at lower cost than Sonnet. |
17
-
| 3 |**Claude Haiku 4.5**| Fast and cost-effective for routine tasks, quick iterations, and high-volume automation. Best for speed-sensitive workflows. |
18
-
| 4 |**Droid Core (GLM-4.6)**| Open-source model with 0.25× token multiplier. Lightning-fast and budget-friendly for automation, bulk edits, and air-gapped environments. |
19
-
| 5 |**GPT-5**| Strong generalist from OpenAI. Choose when you prefer OpenAI ergonomics or need specific GPT features. |
20
-
| 6 |**Claude Opus 4.1**| Highest capability for extremely complex work. Use when you need maximum reasoning power for critical architecture decisions or tough problems. |
|**Deep planning, architecture reviews, ambiguous product specs**| Start with **Sonnet 4.5** for strong reasoning at practical cost. Use **GPT-5 Codex**for faster iteration or **Haiku 4.5** for lighter tasks. |
34
-
|**Full-feature development, large refactors**|**Sonnet 4.5**is the recommended daily driver. Try **GPT-5Codex** when you want faster loops or **Droid Core**for high-volume work. |
35
-
|**Repeatable edits, summarization, boilerplate generation**|**Haiku 4.5** or **Droid Core** for speed and cost savings. **GPT-5** or **Sonnet 4.5** when you need higher quality. |
36
-
|**CI/CD or automation loops**| Favor **Haiku 4.5** or **Droid Core** for predictable throughput at lowcost. Use **Sonnet 4.5** or **Codex**for complex automation. |
37
-
|**High-volume automation, frequent quick turns**|**Haiku 4.5** for speedy feedback loops. **Droid Core** when cost is critical or you need air-gapped deployment.|
36
+
|**Deep planning, architecture reviews, ambiguous product specs**| Start with **Opus 4.5 (default)** for depth and safety. Use **Sonnet 4.5**when you want balanced cost/quality, or **Codex/Codex-Max** for faster iteration with reasoning. |
37
+
|**Full-feature development, large refactors**|**Opus 4.5**for default depth and safety. **GPT-5.1-Codex-Max** when you need speed plus **Extra High**reasoning; **Sonnet 4.5** for balanced loops.|
38
+
|**Repeatable edits, summarization, boilerplate generation**|**Haiku 4.5** or **Droid Core** for speed and cost. **GPT-5.1 / GPT-5.1-Codex** when you need higher quality or structured outputs.|
39
+
|**CI/CD or automation loops**| Favor **Haiku 4.5** or **Droid Core** for predictable, low-cost throughput. Use **Codex** or **Codex-Max**when automation needs stronger reasoning.|
40
+
|**High-volume automation, frequent quick turns**|**Haiku 4.5** for speedy feedback. **Droid Core** when cost is critical or you need air-gapped deployment. |
38
41
39
42
<Tip>
40
43
**Claude Opus 4.1** remains available for extremely complex architecture decisions or critical work where you need maximum reasoning capability. Most tasks don't require Opus-level power—start with Sonnet 4.5 and escalate only if needed.
@@ -47,17 +50,22 @@ Tip: you can swap models mid-session with `/model` or by toggling in the setting
47
50
## 3 · Switching models mid-session
48
51
49
52
- Use `/model` (or **Shift+Tab → Settings → Model**) to swap without losing your chat history.
50
-
- If you change providers (e.g. Anthropc to OpenAI), the CLI converts the session transcript between Anthropic and OpenAI formats. The translation is lossy—provider-specific metadata is dropped—but we have not seen accuracy regressions in practice.
53
+
- If you change providers (e.g. Anthropic to OpenAI), the CLI converts the session transcript between Anthropic and OpenAI formats. The translation is lossy—provider-specific metadata is dropped—but we have not seen accuracy regressions in practice.
51
54
- For the best context continuity, switch models at natural milestones: after a commit, once a PR lands, or when you abandon a failed approach and reset the plan.
52
55
- If you flip back and forth rapidly, expect the assistant to spend a turn re-grounding itself; consider summarizing recent progress when you switch.
53
56
54
57
---
55
58
56
59
## 4 · Reasoning effort settings
57
60
58
-
- Anthropic models (Opus/Sonnet/Haiku) show modest gains between Low and High.
59
-
- GPT models respond much more to higher reasoning effort—bumping **GPT-5** or **GPT-5 Codex** to **High** can materially improve planning and debugging.
60
-
- Reasoning effort increases latency and cost, so start Low for simple work and escalate when you need more depth.
61
+
-**Opus / Sonnet / Haiku**: Off / Low / Medium / High (default: Off)
62
+
-**GPT-5.1**: None / Low / Medium / High (default: None)
63
+
-**GPT-5.1-Codex**: Low / Medium / High (default: Medium)
64
+
-**GPT-5.1-Codex-Max**: Low / Medium / High / **Extra High** (default: Medium)
65
+
-**Gemini 3 Pro**: Low / High (default: High)
66
+
-**Droid Core (GLM-4.6)**: None only (default: None; no image support)
67
+
68
+
Reasoning effort increases latency and cost—start low for simple work and escalate as needed. **Extra High** is only available on GPT-5.1-Codex-Max.
61
69
62
70
<Tip>
63
71
Change reasoning effort from `/model` → **Reasoning effort**, or via the
0 commit comments