DeepSeek vs. Microsoft Copilot vs. Grok: A Quick Comparison with Text Selection InsightsI tested three AI tools—DeepSeek, Microsoft Copilot, and Grok—for their usability in various tasks, including a focus on text selection features. Here’s my concise take:DeepSeek
Pros: Excels in coding, free R1 model, highly accurate, no usage limits.
Cons: Clunky UI, slow responses, frequent server crashes, no image generation. Text selection is cumbersome—long-pressing brings up a menu where you must choose to select text, then a pop-up appears, requiring you to find and reselect the text before copying.Microsoft Copilot
Pros: Free image generation, stable (no crashes), decent coding, unlimited questions, integrates seamlessly with Microsoft Office tools.
Cons: Not as strong as DeepSeek for coding. Text selection mirrors DeepSeek’s process: long-pressing opens a menu, and after choosing to select text, a pop-up forces you to locate and reselect the text to copy.Grok
Pros: I found Grok’s text selection more intuitive—long-pressing a few words in the conversation lets you select them directly without extra steps.
Cons: Still testing, but it has usage limits (10 prompts/2 hours, 3 image analyses/day on the free tier).Verdict: DeepSeek leads for coding despite its flaws. Microsoft Copilot offers reliability and productivity features, especially for Microsoft users. Grok’s seamless text selection stands out, but its limits are a drawback. What’s your experience with these tools?
Ps rewriten my grok and I hate the fact that it threw on information that I didn’t tell it unlike the other ones will not do that unless you prompt it to add more info which in my opinion is much better
Ignore all the extras add in
Did you ever try github copilot for coding?
Why specifically those three?
What about ChatGPT, Claude, and Gemini?
Claude seems to be the best coder.
Copilot is just a dumbed down version of ChatGPT.
I never used Deepseek, I’m not looking to hand my data over to Xi Jinping.
1 Like
I mostly use ChatGPT, they always have the newest innovations first.
1 Like
These are the ones I tested
Great, now that you’ve tested the knockoffs and seen that they’re garbage, you can try the real ones, and see how they compare against Grok.
According to Perplexity
(Ben Shapiro’s personal favorite)
Intelligence and Coding Ability: Claude, ChatGPT, Gemini, Grok, DeepSeek
Which are the “Smartest”?
All the technical details
-
Claude (Anthropic):
- Claude 4 (Opus and Sonnet) is widely recognized as the most advanced in reasoning, context handling, and in-depth analysis, especially for long-form content and complex problem-solving[2][6].
- It features a massive context window (up to 100K tokens), allowing it to process entire books or large codebases in one go, and demonstrates sustained performance on complex, long-running tasks and agent workflows[2][6].
- Regularly leads in coding and reasoning benchmarks, and is considered best-in-class for memory and context retention[2][6].
-
ChatGPT (OpenAI):
- GPT-4 is a strong generalist, excelling in structured writing, coding, and research tasks, with reliable performance across a wide range of domains[6].
- It is highly accessible and versatile, making it a top choice for everyday users and professionals alike[6].
-
Gemini (Google):
- Excels in multi-step reasoning, creative tasks, and multimodal capabilities (text, images, video), and is tightly integrated with the Google ecosystem[6].
- Produces clean, well-structured code and offers multiple response drafts, but may lose conversational context more easily than Claude or ChatGPT[6].
-
DeepSeek:
- Stands out for technical users, particularly in coding and math, and sometimes outperforms GPT-4 in competitive programming challenges[6].
- Fast and efficient, but less refined in general knowledge and writing style[6].
-
Grok (xAI):
- Known for real-time information integration and witty, informal responses, but does not consistently match the accuracy or depth of GPT-4, Gemini, or Claude in structured tasks[6].
- Still developing as a professional tool, with strengths in live data and entertainment[6].
Which are the Best Coders?
| Model |
Coding Strengths |
Notable Weaknesses/Notes |
| Claude |
Currently the top performer in coding benchmarks (e.g., SWE-bench, TAU-bench), excels at complex, large-scale, and agentic coding tasks, and offers deep project context integration (Claude Code)[2][4][5][6]. |
None significant; leads in both benchmarks and real-world use[2][5][6]. |
| DeepSeek |
Highly capable in coding and math, sometimes outperforming GPT-4 in competitive programming; precise and efficient[6]. |
Less creative, more factual; security concerns for sensitive code[6]. |
| ChatGPT |
GPT-4 is highly capable in coding, reliably producing correct solutions and explanations; strong for debugging and brainstorming[6]. |
May struggle with very complex logic or ambiguous prompts[6]. |
| Gemini |
Produces clean, well-structured code; strong in creative and multimodal tasks; good for research and integration with Google tools[6]. |
Loses context more easily in long coding sessions[6]. |
| Grok |
Good for coding and research tasks, especially those needing live data; fast and witty[6]. |
Inconsistent in structured or complex coding tasks; not as mature as others[6]. |
Summary Table: Overall “Smartness” and Coding Ability
| Model |
Smartest (Reasoning/Context) |
Best Coding Ability |
Best Use Cases |
| Claude |
 |
|
Large-scale coding, deep analysis, long context |
| DeepSeek |
|
|
Competitive programming, math-heavy coding |
| ChatGPT |
|
|
General coding, writing, research |
| Gemini |
|
|
Creative, multimodal, Google integration |
| Grok |
|
|
Real-time info, casual coding, entertainment |
Key Takeaways
- Claude 4 (Opus/Sonnet) is currently the smartest and best coding model, dominating both benchmarks and practical use for complex, large-scale, and agentic coding tasks[2][4][5][6].
- DeepSeek is a top choice for technical users needing raw coding and math power, sometimes beating GPT-4 in programming contests[6].
- ChatGPT (GPT-4) remains a strong all-rounder for coding and general smartness, with high reliability and versatility[6].
- Gemini is excellent for creative, multimodal, and Google-integrated tasks, with strong but slightly less consistent coding performance[6].
- Grok is promising for real-time data and informal tasks but is less consistent in structured coding and reasoning[6].
For coding and advanced intelligence, Claude 4 leads the field, followed by DeepSeek and ChatGPT, with Gemini and Grok offering unique strengths in specific areas.
Citations
[1] Claude Code overview - Anthropic
[2] Introducing Claude 4 \ Anthropic
[3] https://www.reddit.com/r/ClaudeAI/comments/1enle9c/can_someone_explain_how_to_actually_use_claude/
[4] Introducing Claude 4 in Amazon Bedrock, the most powerful models for coding from Anthropic | AWS News Blog
[5] https://venturebeat.com/ai/anthropics-stealth-enterprise-coup-how-claude-3-7-is-becoming-the-coding-agent-of-choice/
[6] Grok 3 vs ChatGPT vs DeepSeek vs Claude vs Gemini
[7] https://www.youtube.com/watch?v=vH2f7cjXjKI
[8] https://redblink.com/llm-comparison-chatgpt-gemini-grok-claude-deepseek/
The question wasn’t what is a gpt’s favorite rather it was what is the humans favorite
Because Perplexity was totally trying out each one, forming its own opinions, and then deciding that it likes Claudes personality the best.
My experience with grok so far is terrible… Sometimes doesn’t load well and is giving me limits but on the other hand it has the quickest image generation I’ve seen so far (but copilot is more accurate)
No, he just gets paid by them to advertise.
2 Likes
Not a little bit either. To get mentioned several times every show probably costs them 7 figures per month.
3 Likes
