1. Kimi
Model tested: Kimi K3.1 Pro
Moonshot AI's assistant with agents, deep research and slides, on a free tier and four paid tiers.
Plans
- Adagio: Free
- Plus: $180 per year
- Pro: $372 per year
- Max: $948 per year
If you want a ChatGPT alternative, this list ranks 9 AI chat assistants we tested ourselves. Each one shows its NS-ID Indonesian score, how many of our everyday chat prompts it passed, and plan prices copied from the official pricing page with the date we checked it.
Tested on October 25, 2026 · Prices checked October 25, 2026
Short answer
Our top tier of ChatGPT alternatives right now is Kimi, Perplexity, Claude, Grok, Qwen Chat and DeepSeek: their NS-ID Indonesian scores from the run on October 25, 2026 are within 3 points of each other, which is within normal run to run variation, so we list them by entry price. ChatGPT itself scored 96.4 and passed 8 of 10 test-lab chat prompts; 9 of the 9 alternatives have a free plan.
Swipe to see more columns
| Tier | Tool | NS-ID score | Test lab | Free plan | Cheapest paid plan |
|---|---|---|---|---|---|
| 1 | KimiKimi K3.1 Pro | 99.3 | 10 of 10 | Yes | Plus, $180 per year |
| 1 | GrokGrok 4.7 | 98.0 | 10 of 10 | Yes | Plus, IDR 139,000 per month |
| 1 | ClaudeClaude Sonnet 5.5 | 97.6 | 10 of 10 | Yes | Pro, $20 per month |
| 1 | Qwen ChatQwen3.8 Max (0902) | 97.6 | 10 of 10 | Yes | Not stated |
| 1 | PerplexityPerplexity Sonar | 97.1 | 8 of 10 | Yes | Pro (annual), $17 per month |
| 1 | DeepSeekDeepSeek V4 Pro | 96.8 | 10 of 10 | Yes | Not stated |
| 2 | Google Gemini (Nano Banana)Gemini 3.8 Flash | 93.0 | 10 of 10 | Yes | Google AI Plus, IDR 75,000 per month |
| 3 | Mistral Vibe (formerly Le Chat)Mistral Medium 3.5 | 89.6 | 10 of 10 | Yes | Pro (Education), $5.99 per month |
| 3 | Meta AILlama 4 Maverick | 88.2 | 10 of 10 | Yes | Not stated |
Only tools that are listed in our directory, have a price checked on the official page in the last 90 days, and have results from the same NS-ID suite version (NSID-V1) and our test-lab chat prompts are ranked. Scores are dated test results of the API route named, not a judgement of the company.
Model tested: Kimi K3.1 Pro
Moonshot AI's assistant with agents, deep research and slides, on a free tier and four paid tiers.
Plans
Model tested: Grok 4.7
xAI's assistant with live web and X search, plus image and video generation.
Plans
Model tested: Claude Sonnet 5.5
Anthropic's AI assistant for chat, writing, files and coding, with Free, Pro and Max plans.
Plans
Model tested: Qwen3.8 Max (0902)
Free chat assistant from Alibaba's Qwen team with search, deep research and media generation.
Model tested: Perplexity Sonar
AI answer engine with cited sources and a choice of models from several labs.
Plans
Model tested: DeepSeek V4 Pro
Free chat assistant from DeepSeek with DeepThink and Search modes.
Model tested: Gemini 3.8 Flash
Google AI assistant whose app generates and edits images with the Nano Banana models.
Plans
Model tested: Mistral Medium 3.5
Mistral AI's assistant for chat, search and coding, renamed from Le Chat to Vibe.
Plans
Model tested: Llama 4 Maverick
Meta's assistant on the web, in its own app, and inside WhatsApp, Instagram, Messenger and Facebook.
ChatGPT
General AI chat assistant from OpenAI with Free, Go, Plus and Pro plans.
Swipe to see more columns
| Plan | Price | Source |
|---|---|---|
| Free | Free | Source:chatgpt.com, checked October 25, 2026 |
| Go | IDR 75,000 per month | Source:chatgpt.com, checked October 25, 2026 |
| Plus | IDR 349,000 per month | Source:chatgpt.com, checked October 25, 2026 |
| Pro | IDR 1,889,000 per month | Source:chatgpt.com, checked October 25, 2026 |
These tools are listed in the directory but have not been through our tests yet, so they are not ranked.
Among the tools we tested that have a free plan, Kimi had the highest NS-ID score (99.3) on October 25, 2026. Free plan limits change often, so check the plan list on each card.
Kimi, Perplexity, Claude, Grok, Qwen Chat and DeepSeek formed the top tier in our NSID-V1 run (within 3 points of the best NS-ID total, 99.3 out of 100). NS-ID covers comprehension, formal writing, tone and register, translation, local knowledge and following instructions.
No. We call the model behind or closest to each app through its API, with fixed prompts and saved outputs. The app version can behave differently because of its own system prompt, search and memory features.
Every price on this page shows the date we copied it from the official pricing page. Anything older than 90 days is marked as needing a re-check and drops out of the ranking.