Best AI Models for Chat
Ranking of language models for conversational assistants and chatbots — helpfulness, response speed and cost at scale.Updated August 2026. Based on independent benchmark data.
Why This Matters
Choosing the right model for chat is about more than raw benchmark scores. The models below are ranked using a weighted blend of category-specific benchmarks, arena Elo, and the intelligence index — but real-world performance depends on your context window, latency budget, and integration path.
For chat, we weighted the benchmarks that most strongly predict success in production: task accuracy first, then speed and cost efficiency. Leaders on this page consistently outperform on both standardized tests and real-world workloads.
If you operate under a strict budget, check the Best Value LLMs ranking. If latency matters more than raw quality, see Fastest LLMs. Both pages use the same underlying dataset with different weightings.
| # | Model | Provider | Score | Speed | Price (output) |
|---|---|---|---|---|---|
| 1 | GLM-5.3 Flash | Z.ai | 80.1 | 182 t/s | $0.3/M |
| 2 | Qwen3.8 Flash Next | Alibaba | 79.6 | 168 t/s | $0.4/M |
| 3 | Gemini 3.1 Pro | 76.8 | 136 t/s | $12/M | |
| 4 | HyperNova 60B | Multiverse | 75.1 | 367 t/s | $0.14/M |
| 5 | Gemini 3.8 Flash | 73.3 | 302 t/s | $3.75/M | |
| 6 | Gemini 3.8 Flash (Medium Reasoning) | 73.3 | 315 t/s | $3.75/M | |
| 7 | Gemini 3.8 Flash (Low Reasoning) | 73.3 | 340 t/s | $3.75/M | |
| 8 | GPT-6 Sol (Non-reasoning) | OpenAI | 72.9 | 265 t/s | $4/M |
| 9 | GPT-6 Sol (xhigh) | OpenAI | 70.5 | 210 t/s | $4/M |
| 10 | GPT-6 Sol | OpenAI | 70.5 | 220 t/s | $4/M |
| 11 | GPT-6 Sol (high) | OpenAI | 70.5 | 226 t/s | $4/M |
| 12 | GPT-6 Sol (medium) | OpenAI | 70.5 | 238 t/s | $4/M |
| 13 | GPT-6 Sol (low) | OpenAI | 70.5 | 250 t/s | $4/M |
| 14 | Granite 4.2 3B Instruct | IBM | 69.6 | 148 t/s | $0.2/M |
| 15 | Qwen3.7 Max | Alibaba | 69.1 | 203 t/s | $7.5/M |
| 16 | LFM2.5-8B-A1B | Liquid AI | 68.1 | 338 t/s | $2/M |
| 17 | LFM2.5-VL-1.6B | Liquid AI | 68.1 | 395 t/s | $2/M |
| 18 | DeepSeek V4.1 Flash (Non-reasoning) | DeepSeek | 66.9 | 240 t/s | $1.2/M |
| 19 | DeepSeek V4.1 Flash | DeepSeek | 66.4 | 214 t/s | $1.2/M |
| 20 | DeepSeek V4 Flash 0420 (Non-reasoning) | DeepSeek | 66.0 | 210 t/s | $1.1/M |
Top 5 Chat — Quick Comparison
Methodology
Rankings are based on independent benchmark evaluations. The chat score is computed from multiple standardized benchmarks and normalized to 0–100. Data is updated regularly from public benchmark results.
Related Rankings
Frequently Asked Questions
Which AI model is best for building a chatbot?
For chatbots, balance conversational quality with speed and cost — flagship knowledge plus high token-per-second throughput wins.